Chronic measurement of the stimulation selectivity of the flat interface nerve From an electrophysiological standpoint, the FINE is a viable alternative for. Nerve-fiber recording and stimulation from the nervous system have .. ) have been developed recently to demonstrate its viability for. Stimulation Selectivity of the “Thin-Film Longitudinal Intrafascicular Electrode” In conclusion, TIME is a viable neural interface for selective activation of multiple .
stimulation of and viability Selectivity
We used ImageJ to convert the before image to also be 8-bit grayscale, then combined each pair of images into a stack, preserving scale information from the original file. Next the before stimulation image was subtracted from the after stimulation, and all such difference images were averaged together.
This created a heat map for spot stimulation on each type of silicon and for whole field stimulation on untreated silicon. We first assessed the ability of rat primary cortical neurons to be cultured on a silicon-based substrate by immunostaining and electrophysiological techniques.
The silicon based substrate, polished silicon, was coated with poly-lysine and rat primary cortical neurons were plated as previously described Arikkath et al. Neurons were fixed and immuostained with antibodies to neuronal and synaptic markers Figure 1A and transfected with EGFP to assess neuronal morphology and subjected to analysis by confocal microscopy. Neurons had normal expected morphology, expressed the dendritic marker, Map2, and axonal marker, Tau, and formed synapses as indicated by staining with excitatory and inhibitory synaptic markers, Gad65 and vGlut1.
Spines also appeared to have the expected morphology Figure 1B. In addition, we also performed electrophysiological recordings from neurons cultured on the polished silicon substrate to test the normal occurrence of spontaneous EPSCs and action potentials to assess neuronal health on this substrate Figure 2.
Neurons cultured on the silicon chip were morphologically and electrophysiologically indistinguishable from primary neurons cultured on standard glass coverslips. Thus, the silicon based substrate is a highly permissive substrate for primary rat cortical neurons. Neurons cultured on silicon chip retain morphology and express neuronal and synaptic markers. A Confocal images of cortical neurons grown on polished silicon chip and immunostained with dendritic Map2 , axonal Tau and synaptic Gad65 and vGlut1 markers at DIV 7 and B Confocal images of cortical neurons grown on the polished silicon chip and transfected with GFP.
Note that neuronal morphology is well preserved and normal morphology of spines is observed. Scale bar 20 and 2 um. Neurons cultured on silicon chip exhibit typical electrophysiological properties. Data were filtered at 2 kHz 1 kHz for presentation and digitized at 20 kHz. B Whole-cell current-clamp recording. Action potential in response to current injection. We then designed a chamber that would allow photoconductive stimulation of neurons cultured on the silicon based substrate using an inverted confocal microscope Figures 3A—D.
The inverted configuration is the configuration of choice in labs that image live neurons by light microscopy.
Photoconductive stimulation uses light to lower the electrical resistance of a section of the substrate, allowing a voltage pulse applied to the back plane of the substrate to selectively shunt through the illuminated region and stimulate the cells in the near vicinity of the shunt.
Unlike conventional glass and plastic substrates, silicon is opaque, which presents a challenge to inverted imaging where cells are usually imaged through their substrates. To solve this issue, our chamber holds the silicon substrate with its adherent cells suspended a small distance above a glass coverslip bottom.
The technique can be applied either by pulsing the voltage at a constant light flux or by pulsing the laser while holding the sub-threshold voltage constant. The cell side and back side of the silicon chip are exposed to isolated baths and voltage pulses are delivered from an external stimulator to electrodes in these baths. Design and Set up of Photoconductive Stimulation Chamber for inverted microscope. A Chamber is comprised of a base, silicon chip, insert, and clamps.
The bottom of base chamber is a glass coverslip. Feet on coverslip keep silicon chip elevated, allowing fluid access to neurons cultured on chip. The chip is pressed onto feet by rubber gasket on insert, which is held down by clamps. This forms two chambers, in base and insert, electrically isolated by the silicon chip. Each chamber has a platinum electrode. B 3D blown-up cut away of chamber.
C Chamber setup on scope. Electrodes are connected to a Grass S48 stimulator. D Photograph of chamber during experiment. We first examined the frequency dependent responses of neurons cultured on conventional glass coverslips in a glass bottom chamber to electrical field stimulation by means of external electrodes. Neurons were loaded with the calcium indicator dye, Fluo-4 Taylor et al.
As expected, the response of the cells as indicated by an increase in fluorescence intensity was tightly correlated with the frequency of stimulation. Calcium imaging frequency response of cultured rat cortical neurons with field stimulation or photostimulation. A Fluo-4 calcium imaging intensity of cultured cortical neurons in response to 5 s of 1 Hz B 5 Hz and C 10 Hz field stimulation. Rainbow colormap images of fluo-4 based calcium response from cell undergoing 1Hz photoconductive stimulation on H silicon chip L porous oxidized silicon or P porous carbonized silicon.
Error bars show standard error. To examine the ability of neurons to be stimulated using photoconductive stimulation, we chose the approach of pulsing the applied voltage while maintaining a constant light flux. Using a similar Fluo-4 dye loading approach, we examined the ability of neurons cultured on the polished silicon substrate to elicit cellular responses in a frequency dependent manner in response to a combination of constant laser light and electrical stimulation of the silicon substrate at 1, 5, and 10 Hz Figures 4E—H , supplementary video 2 in comparison to the similar responses elicited by field stimulation.
We similarly characterized neuronal responses in neurons cultured on the silicon wafers that contained a surface texturing of mesoporous silicon oxide Figures 4I—L , supplementary video 3 and mesoporous carbonized silicon Figures 4M—P , supplementary video 4. Stimulation was targeted by using the digital zoom factor to restrict laser scanning to a single cell body.
Photoconduction was induced by using a second laser at high power concurrently with the imaging laser. Neurons responded to stimulation with oscillatory increases in Fluo-4 fluorescence at the stimulation frequency.
These response patterns were similar between all three silicon substrates and similar to those observed in neurons that were stimulated using field stimulation. Thus, the cell response is synchronized with the frequency of stimulation with the photoconductive stimulation technique, similar to that observed with field stimulation. One highly desired characteristic of a neuronal stimulation technique is the ability to stimulate neurons in a spatially controlled manner.
We measured the ability of the technique to stimulate cells in a spatially confined manner, by manipulating the laser during stimulation.
We loaded neurons cultured on the 3 different silicon substrates with Fluo-4, then imaged before and after a 5 s, 10 Hz stimulation train. We used the spot scan function to target the laser at a single cell body and assessed our ability to selectively stimulate that cell.
We were usually able to stimulate a single targeted cell on untreated polished silicon, while similar stimulation on the porous oxidized silicon and porous carbonized silicon Figures 5A—F resulted in a larger region of stimulation. Spatial resolution of photoconductive stimulation of rat primary cortical neurons on three different substrates.
Representative set of images showing targeted neuronal excitation visualized by Fluo-4 calcium imaging of neurons cultured on A,B polished silicon, C,D porous oxidized silicon or E,F porous carbonized silicon. G Fluorescence increase as a function of distance to stimulation point and H percentage of cells activated as a function of distance from stimulation point for neurons cultured on silicon blue squares , porous oxidized silicon green triangles and porous carbonized silicon red diamonds.
Error bars show standard error from independent experiments. I—L Rainbow Heat map representing intensity before and after 5 s of 10 Hz stimulation. L Average heat map for neurons cultured on silicon chip with entire field stimulation.
Neurons demonstrated robust fluorescence increase in response to stimulation on all three substrates Figures 5A—F,G. By altering the area scanned by the laser, a larger region of the polished silicon chip can be activated Figure 5L.
Thus, the spatial selectivity of the stimulated region can be altered by use of the desired silicon based substrate or manipulating the area scanned by the laser. We have designed and implemented a chamber for photoconductive stimulation using a regular inverted confocal microscope. We have demonstrated the applicability of a non-invasive photoconductive stimulation for primary neurons in culture using a laser based photostimulation.
More importantly, we have demonstrated that this technique can be spatially manipulated easily by the choice of silicon substrate or altering the area illuminated by the laser. The chamber and technique can be easily implemented on any standard inverted microscope.
The system is conducive to both acute and long term studies that that couple neuronal photoconductive stimulation with time-lapse imaging. The same chamber, technique, and principles can be applied to any excitable cell type, including other types of neurons, neuron-astroctye co-cultures or cardiomyocytes. By combining this technique with other existing technologies, the system can perform cell biological studies that address neuronal network dynamics and dynamic cell biological studies that assess cellular alterations in response to synaptic activity by molecular fluorescent indicators or translocation of neuronal proteins in response to neuronal stimuli.
Our data indicate that silicon based substrates are highly permissive for neuronal growth and photoconductive stimulation. Scaling up of the technique and adapting to multiwall formats would be important advances for drug discovery and characterization based screening techniques. Similarly, by patterning or interspersing the silicon based substrates on glass surfaces, further spatial control over stimulation can be achieved.
Thus, this technique is a simple but powerful, versatile and easy to use tool that can be easily implemented in a standard live chamber on an inverted scope with little specialized equipment and is likely to find broad application in cellular and molecular neuroscience. Jacob Campbell and Jyothi Arikkath designed experiments, Jacob Campbell performed all photoconductive experiments, Dipika Singh and Jyothi Arikkath devised the technique for freezing primary neurons, Dipika Singh and Jyothi Arikkath characterized the frozen primary neurons, Jyothi Arikkath performed confocal imaging analysis for immunostaining and transfected neurons, Geoffrey Hollett participated in generating silicon substrates under direct supervision of Michael J.
Dravid performed the electrophysiology experiments. Jyothi Arikkath and Jacob Campbell drafted the manuscript with input from the other authors. Jyothi Arikkath supervised the whole project. All animal experiments were conducted in compliance with approved UNMC protocols. Arikkath and Dipika Singh are the creators of the freezing media for the primary neurons used in this study and receive royalties from sales of the freezing media.
All other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at: PICK1 loss of function occludes homeostatic synaptic scaling. Presynaptic neurexin-3 alternative splicing trans-synaptically controls postsynaptic AMPA receptor trafficking. Delta-catenin regulates spine and synapse morphogenesis and function in hippocampal neurons during development.
Tracking axonal action potential propagation on a high-density microelectrode array across hundreds of sites. Culturing pyramidal neurons from the early postnatal mouse hippocampus and cortex. The activity-dependent transcription factor NPAS4 regulates domain-specific inhibition. Remodeling of synaptic actin induced by photoconductive stimulation.
Activity-dependent neuronal signalling and autism spectrum disorder. Photoconductive stimulation of neurons cultured on silicon wafers. LTP requires a reserve pool of glutamate receptors independent of subunit type. Nature , — doi: Activity-dependent relocation of the axon initial segment fine-tunes neuronal excitability.
Activity-driven mobilization of post-synaptic proteins. Encoding and transducing the synaptic or extrasynaptic origin of NMDA receptor signals to the nucleus. Cell , — doi: A centrosomal CdcAPC pathway controls dendrite morphogenesis in postmitotic neurons. Measurement of synchronous activity by microelectrode arrays uncovers differential effects of sublethal and lethal glutamate concentrations on cortical neurons.
Caspase-3 activation via mitochondria is required for long-term depression and AMPA receptor internalization. Optogenetic inhibition of synaptic release with chromophore-assisted light inactivation CALI. Input-specific spine entry of soma-derived Vesl-1S protein conforms to synaptic tagging.
Development of an in vitro model of neuronal activity induced excitotoxicity using photoconductive stimulation. Cell Calcium 47, — Elfn1 regulates target-specific release probability at CA1-interneuron synapses.
Microfluidic local perfusion chambers for the visualization and manipulation of synapses. Ultrafast endocytosis at mouse hippocampal synapses. This technique may be superior to vibro-tactile substitution in that it reduces the grasping force error of prostheses, thereby providing more accurate spatial sensory information Patterson and Katz, ; Antfolk et al.
However, Antfolk et al. Regarding all the modalities noted above, achieving a selective and stable interface remains a challenge. Hence, researchers have looked for other near-to-natural-sensation modalities, among which peripheral implants are popular these days.
In the next section, we will review the progress made in the neuroprosthetics using implantable interfaces at neuraxis and we will explore peripheral nerve-machine interfaces used for recording and stimulation.
In the indirect methods noted above, complete selectivity and longevity have not yet been achieved. These modalities are still unable to provide selective sensory information in a stable manner. To overcome this issue, the peripheral interface technology has advanced more in recent years, and researchers are now focusing more on invasive technologies Navarro et al. Tyler and Durand focused on the fact that for those amputees who only lost sensory end organs limbs , the peripheral nerves connecting the brain to the lost organs retain the normal functionality.
Thus, through a synthetic activation of these pathways, perception of sensation can be achieved Micera and Navarro, There are several locations at which the interaction with the residual somatosensory system can be established Weber et al. Direct stimulation on the nerve has earned extensive popularity in providing sensory feedback to prostheses. Preclinical works on the approaches related to the brain Bensmaia and Miller, , dorsal root ganglion Weber et al.
In the twenty first century, two important developments have transformed the field of neuroprosthetics. The first is the improvement in bionic arms or prosthetic limbs that can replicate the functions of a natural human arm. The second is the enhancement of algorithms that decode the intended movements of an amputee from neuronal activities in the motor cortex area of the brain. By combining these innovations, it is now possible for a human patient to perform tasks with a bionic arm by thoughts alone Collinger et al.
Most existing interfaces are either from the nerve or the cortex Tabot et al. The restoration of somatosensation, either from cortical stimulation or peripheral-nerve stimulation, is required for effective bidirectional communication and a sense of feeling or embodiment Dornfeld et al. The essential need for touch in everyday life has led several research groups to develop different techniques for its artificial restoration. The necessary methodology entails stimulation of the peripheral nerve or a somatosensory area of the brain S1 with trains of electrical pulses to evoke percepts that transmit information from the grasping object Downey et al.
Amputees or patients that use a bionic limb controlled through such a bidirectional peripheral-nerve-interface perceive the prosthesis as an integral part of themselves rather than as a piece of hardware attached to their arm Marasco et al.
The dexterity of a bionic arm can be improved through the restoration of somatosensation by stimulating the peripheral nerve, because in some cases such a manipulation of an object, visual feedback is considered to be a poor substitute Bensmaia, To obtain such a feedback, several peripheral-nerve-interface approaches have been devised over the past two decades. As noted above, electrical interfaces can be established anywhere in the cortex of the brain to the end organ.
Sometimes interfaces are penetrative and sometimes stimulation can be given externally. In a broad spectrum, evaluation of these interfaces can be based on selectivity and longevity: Longevity is the measure of stable interaction of the electrodes with the same population of sensory afferents over time, and selectivity is the measure of its interaction with specific parts of sensory afferents.
For the evaluation purpose, computational models of tactile afferents also have been deployed, which can simulate a population of afferents, in real time, in milliseconds and with precision Kim et al.
Furthermore, stability can be defined as the duration that the information related with the activity measured by the electrode should remain constant over the life of the interface.
Greater stability of a stimulating electrode results in a more natural feeling of an artificial touch and is also important in achieving longevity for sensory feedback. Meanwhile, a higher selectivity can be achieved through intra-neural implants and, along with that, stability over a longer period also can be achieved.
To increase selectivity, penetrating array type interfaces are used. However, extra-neural implants might stimulate a population of axons. The fundamental problem in cortical interfaces is longevity Warren et al. The neurons or neuronal tissues at the implanted surface and the electrodes implanted interface degrade over time McCreery et al. These changes can affect the stimulation and recording abilities of the electrodes for sensory feedback or decoding the attempted movements Perge et al.
Peripheral nerve stimulation through different interfaces is the hot issue for restoring sensations. In the next section, we will explain the recent advances for peripheral-nerve interfaces. The functional properties, selectivity, and biocompatibility will be discussed along with the advantages and disadvantages of various peripheral-nerve interfaces. The techniques in wide use for restoration of sensory feedback through electrical stimulation will be highlighted, though some other potential stimulation techniques such as the optogenetic modality targeted neural signaling Towne et al.
In general, two broad categories of peripheral-nerve-interfaces are extra-neural and intra-neural. The extra-neural or extra-fascicular interface with a circular shape that surrounds the peripheral-nerve is a noninvasive method to the nerve itself. The electrodes do not penetrate the protective sheath perineurium , and thus are less invasive and minimize the disturbances to the neural tissue. The cuff interface is most common among the extra-fascicular types.
The helical Agnew et al. Their circular shape offers a disadvantage of minimal interaction with neural tissues. Tyler and Durand introduced the Flat Interface Nerve Electrode FINE , which is another form of extra-neural interface that was developed to increase the surface area without penetrating inside the nerve and to maintain the naturalistic shape of the nerve. Its recording ability Yoo and Durand, and selectivity of the periphery of the stimulated nerve have been demonstrated Schiefer et al.
Since the electrodes and the nerve fibers are separated by the perineurium, higher stimulation currents are required to achieve sensations than the case of intra-fascicular interfaces Grinberg et al. Leventhal and Durand have employed a higher value of current resulting in activation at subfascicle level and achieved a low selectivity, and also their results were not repeatable: Hence, a similar tactile perception, which is not a naturalistic pattern of neuronal activation, was evoked.
During grasping with an intact hand, every afferent fiber perceived differently and responded as per the object' features that invaded the fiber's receptive field, while electrical stimulation through the electrode sometimes produced a highly unnatural feeling or paresthesia due to synchronous activation of a large population of axons.
The temporal modulation of stimulation pulse trains somewhat mitigated the effect of tingling and paresthesia.
A stable and selective configuration of FINE and spiral interfaces has been achieved in clinical trials for more than 3 years for individuals with limb loss Tan et al. Selectivity also has been achieved in this way, but only on the periphery of the nerve, not up to the axonal level.
The main drawback of extra-neural interfaces is their low selectivity. In fact, being wrapped around the nerve, the electrode records the whole electrical activity of the nerve. As noted above, in order to reach afferent axons from the periphery of a nerve, they must provide high stimulation currents as compared to the intra-neural ones.
For these reasons, an intra-fascicular type that penetrates the nerve itself has been introduced. As the name suggests, an intra-neural interface penetrates the protective sheaths. The least invasive designs such as the groove interface Koole et al.
These interfaces physically insert electrical contacts within the nerve. These interfaces, unlike extra-neural ones, tend to have more contacts within the peripheral nerve.
LIFE has some drawbacks that include i a fixed distance between the electrode and the nerve which limits its selectivity, ii the interface stiffness causes micro-movements that, in the long term, may damage the nerve.
It has been fabricated on a micro-patterned polyimide substrate, allowing several contacts in one interface. Its flexible structure allows a better adaptation to the nerve shape preventing damage due to an excessive stiffness. They are less invasive than recent multichannel array type interfaces. Selective stimulation and recording have been achieved through TIME at both intra-fascicular and inter-fascicular levels Badia et al.
Also, multi-electrode-array-based interfaces Byun et al. These interfaces have great benefits for individuals who have suffered a limb loss. This type has varying electrode densities according to several designs proposed in the literature with multiple penetration depths Branner et al. They are inserted perpendicularly into the peripheral nerve leading to a higher risk of nerve damage. They require a lower current for activation of nerve fibers due to their close proximity to axons and fibers; significantly therefore, a small group of fibers can selectively be stimulated Branner et al.
Coordinated grasp and sensory responses by stimulating the peripheral nerves of a monkey was demonstrated with the implantation of USEA. Moreover, with short-term implantation of intra-fascicular electrodes, increasing stimulation thresholds have been observed Boretius et al.
The reported number of sensory perceptions and the locations evoked by intra-neural interfaces are very similar to those of the long-term extra-neural approaches, which led Grill et al. One of their disadvantages is the tendency to damage the nerve by penetration, which reduces the long-term stability.
RPNI consists of an electrode and a residual peripheral nerve, which is neurotized by transacting the nerve and inserting the electrode in between them; it is an internal interface for signal transmission with the external electronics of a prosthetic limb.
Effectively, it is a sieve made of silicon, ceramic or polymer with a large number of fine holes that is placed inside the transected nerve. The increase in growth rate of regeneration of individual or a group of fibers through those holes can activate selective stimulation and make it possible to record action potentials Micera and Navarro, ; Sando et al.
In principle, a high number of selective contacts can be achieved by reducing the size of the sieve along with an increased number of fine holes Lago et al. The neurotized interface serves as a biological amplifier and provides a long-lasting site for implantation.
Studies revealed that the RPNI undergoes robust regeneration, neurotization, and revascularization Urbanchek et al.
Polyimide-based sieve electrodes have been shown to be biocompatible Stieglitz et al. Further experimentation with the interface revealed the stability and formation of new neuromuscular junctions within the muscles for improved function in amputees.
Also they demonstrated the 7-month viability of the RPNI. On the other hand, stimulation of a small number of regenerated fibers was shown feasible using the regenerative electrodes. By matching the regenerated nerve fascicles with the original receptive fields, an adequate feedback might be delivered. By targeting the fiber growth for a specific feedback of perception from a distinct neural population, there might be a possibility of generating a different fiber population that might not be required for the receptive field Lotfi et al.
Sensory feedback has been achieved from tactile and force sensors embedded in the prosthesis by stimulation of appropriate afferent fibers in the transected nerves. Another methodology to attain feedback has been proposed by Nghiem et al. In this method, an insulated electrode placed on the surface of a muscle was used to stimulate the muscle, which then depolarized the afferent nerve within the muscle to provide sensory feedback.
This new design can eliminate the problem of direct nerve stimulation that is inherent in intra-neural interfaces and also limits potential peripheral nerve damage. This sensory RPNI has the ability to transduce distinguishable and graded sensory signals across the peripheral nerve when being stimulated electrically. Although RPNI is still in an early stage of development, this technique has the capability to access sensory pathways and provide stable sensory feedback. A disadvantage of this modality is that the functional use of electrodes might require several months due to nerve regeneration.
Furthermore, the nerve might degenerate with time, which will lead to a substantial loss of stimulation ability of the implanted electrode Lago et al. A number of useful results have been obtained using the RPNI technique in experimental models Ceballos et al.
However, some challenges remain, which hinder its clinical usability. The most important breakthrough of regenerative electrodes will be their implantation in the peripheral nerves of an amputee for bidirectional interfaces. TR is a nerve-machine interface that has been developed to make prosthetic control and feedback more intuitive. This method is considered to be fully neither invasive nor non-invasive; it lies in between due to the surgical requirement for its implementation.
It has demonstrated a success in improving the motor control signals for both transhumeral and shoulder disarticulation amputation Dumanian et al. It has also shown promising to sensory outcomes by using rerouted residual median, radial, and ulnar nerves. This can help the transmission of sensory feedback and the attainment of electromyographic EMG signals by surface electrodes from the reinnervated site to control the prosthetic limb Kung et al. The EMG signals corresponding to the intended movement of a patient are generated from muscle contractions of redirected nerves.
When the target skin of these patients was touched, they felt as if their missing limb was being touched Kuiken et al. Studies of two amputees also have demonstrated a touch perception that aroused in the target skin: The amputees had a strong impression that the sensations arising from stimulation of the reinnervated skin site were projected to their missing limb.
Furthermore, the sensory afferents remained stable for years after surgery. This methodology is being expanded in the field, evoking cutaneous or tactile sensations Hebert et al. Using TR, sensation in the hand as evoked by a reinnervated chest skin along with other senses like touch, temperature, pain, and a limited graded force discrimination have been demonstrated Kuiken et al. There is a possibility of generating tactile feedback, but this technique is not naturalistic in restoring proprioception.
The advantages of TR technique, as stated by Hebert et al. A series of results for patients who have undergone TR are; i non-invasive stimulation at the innervated site resulting in a generation of perceived sensory information a cutaneous sensation in the median nerve of the hand, ii effective detection of touch, pain, temperature, and proprioception to some extent, and iii stable reinnervated area for the detection of graded forces. TR does not only help in providing sensory feedback but also increases the size of the sites that can be used to control the prosthetic hand Kuiken et al.
However, Kristeva-Feige et al. The proximity of the sites of sensory feedback and motor control can be considered as the major disadvantage of the TR technique. Also, the tactor array used for elicitation of sensory feedback at the TR location has only a limited ability in producing a wide range of sensation. The required calibration routine for taking the tactor array on and off has made this option rather difficult for daily use.
Long-term stability and selectivity can be obtained if implanted interfaces are bio-integrated or biocompatible with the peripheral nerve. Foreign body reaction developed around the implants can affect the signal to noise ratio of the recorded motor commands and change the stimulation threshold values for sensory feedback.
Therefore, existing interfaces and their fabrication have been considered as a practical choice for a prosthesis that intends to the recovery of sensorimotor abilities with characterization of long-term usability and biocompatibility. Such interfaces have offered a high-resolution means to access the information from the peripheral nerve by processing the multi-unit recordings and stimulations.
Electrical stimulation of a single or a micro-electrode array has required current pulsations for eliciting action potentials. However, this approach is required to fulfill some conditions; i specifically designed interfaces for peripheral nerves, ii invasive enough to reach a target axon, iii ability to communicating bi-directionally, iv selective and stable electrical interfacing, v minimal tissue damages and influence of foreign body responses, and vi most importantly a biocompatible and biostable interface.
Several types of interfaces proposed in the literature showed different material and chemical properties. Out of which, polymer based neural interface is popular. A conducting polymer is often used as a coating material on electrodes to increase charge-injection capacity for neural stimulation and to get high signal-to-noise ratio SNR for neural recording.
Polypyrrole PPy and poly 3,4-ethylenedioxythiophene, PEDOT are the most widely used polymer coatings for neural electrodes due to their biocompatibility.
Different types of interfaces have been tested by several research groups on animals for checking their biocompatibility, recording and stimulating abilities for instance; Polyimide-based intra-neural implants Wurth et al. As noted in the previous sections, several research groups have tested the capacities of intra- and extra-fascicular interfaces by implantation in the ulnar and median nerves to selectively evoke sensation in amputees through multiple electrodes.
The artificial sensory signals are transmitted to the peripheral nerves by implanting electrodes providing a stimulated current that is proportional to the original input. By means of real-time sensory feedback, the amputee is able to move a prosthetic arm, apply a grip force via the recording of fine motor commands, and also feel a sensation without any audio and visual aids. Some studies have shown such ability of an amputee to differentiate objects according to their perceived characteristics e.
The ability of object recognition and simultaneous encoding of sensory information for manipulation of grip forces are the excitatory developments for amputees. However, its widespread applicability is reduced by the limited number of studies and participants to date. Given the limited long-term data, it is significant to consider the physical conditions of the electrodes implanted in the peripheral nerves along with their capacity to provide a long-term stable interface Yoshida et al.
Structural changes of nerves can occur due to implantation. Fibrosis can hinder the response of an electrode and constantly reduce its performance. Implantation of tf-LIFE 4 in humans has shown a complete termination of sensory detection after 10 days of stimulation, which was due to the foreign-body response of tissue fibrosis Rossini et al. Summary of human neural implant studies and modalities for restoration of sensory feedback. Additionally, some cortical implants like the Utah Electrode Array and the customized multi-channel stimulator for a cortical visual neuroprosthesis can be used for blinds Ferrandez et al.
Such stimulators have been proven to inject current charge in a safe and precise way in an acute animal experimentation. Another implant is the planar multi-electrode array for recording from a rat auditory cortex and visualizing a spatiotemporal structure of the cortical activities Tsytsarev et al.
The key to a further development of neuroprosthetics is to record simultaneous single-unit, multi-unit, and local field potential activity from multiple brain sites.
Most recent study by Pothof et al. These probes have successfully recorded single unit activities up to 26 days from the brain of a monkey that suggests its potential usefulness on human applications. It is evident from above discussions that more research into long-term interfacing implants for bidirectional control of prostheses is required. In the following section, some parameters that can help in selective neural stimulation for elicitation of distinct and graded sensations sensory feedback are discussed.
Spikes are responsible for conveying information in the axons of a nerve, and the spike rate or frequency is involved in transmitting sensorimotor commands in a single axon Tyler, Electrical stimulation in the form of spikes has been given to the residual stump nerve of a patient through typically implanted electrodes, and psychophysical judgments and verbal reports were gathered Polasek et al. Stimulation through different electrodes is one way of manipulating evoked sensory percepts.
Different populations of afferents can be activated through these individual electrodes, which can have distinct receptive field locations Saal and Bensmaia, and different sub-modality compositions of sensory afferents mix of slowly adapting SA , rapidly adapting RA , and pacinian corpuscles PC afferents.
Thus, by evoking sensations through different electrodes at different locations in a phantom or residual limb, the field of projections of active fibers, which is the area where sensation is felt, can be determined. These sensations can also be determined through a sub-modality composition of activated fibers. For instance, if afferent populations of SA, RA, or PC fibers are stimulated, the evoked sensation will be of pressure, flutter, or vibration, respectively Torebjork et al.
Moreover, the pulse shape is of extreme importance. The electrical pulse is transmitted to the nerve fiber using implanted electrodes. The field of pulse magnitude decays as the distance from the implanted electrode increases.
Rattay and Aberham concluded that nerve activation is proportional to the rate of change of the voltage along the axons. Therefore, large, myelinated and closer axons can be activated by square pulses before small, unmyelinated and distant axons.
A ramp pulse can be used to activate distal axons before closer ones Grill and Mortimer, ; Grill et al. In implanted devices, exponential-shaped waveforms, utilizing minimal energy, can be used to minimize the power requirement for activation of larger and closer axons Wongsarnpigoon and Grill, ; Krouchev et al. As explained above, providing stimulation using different electrodes can elicit sensation of several different qualities.
In addition to the closeness of intra-neural interfaces physically, another powerful tool is the design and manipulation of the spatiotemporal electro-magnetic field using extra-neural interfaces. These interfaces, with the help of multiple electrodes Schiefer et al. The standard nerve stimulation paradigm is a train of identical, charge-balanced, square electrical pulses characterized by pulse amplitude, pulse width, and pulse repetition period or inter-pulse interval.
Traditionally, these three parameters are time-invariant and fixed in value. The intensity of a pulse can be manipulated in three different ways by varying pulse width Tan et al. It ranges from the lowest charge that can evoke a sensation all the way up to the value that elicits unnatural sensation or paresthesia. By alteration of these stimulation parameters, sensations of varying quality can be evoked.
In somatosensory applications, intensity is proportional to the variation of the stimulation frequency, and the resulting sensory perceptions can be changed with bursting pulse trains Tan et al. In the patterned frequency paradigm, synchronous activity in a population of axons can be generated by maintaining the strength of the stimulation pulse constant.
However, this cannot help in distinguishing complex sensory information. For generation of a patterned fiber activity, several modalities such as patterned field distribution of stimulation and patterned stimulation intensity have been employed. In each paradigm, non-synchronous activity in a population of axons is generated by creating a shift in the field between stimulation pulses. This information, if controlled properly, can be utilized for the restoration of several different somatosensory percepts.
The nerve stimulation has been shown to induce different types of sensations; for instance, proprioceptive and touch sensations. Complex qualities of touch such as vibration on the skin, tingling, tapping, stinging, brushing, and itch have been observed during experimentation Tan et al.
Stimulation can infrequently evoke proprioceptive sensations, for example, sensations of movement of a finger or joint or a specific hand configuration. Systematic study of percept qualities and gradedness of these sensations have not been completed yet. Data on contact forces between the perceptually available object and the skin are required for grasping and manipulating an object Johansson and Flanagan, , because too much force is likely to damage the prosthesis or the object, while too little force can cause slippage Witney et al.
In previous studies, contact force has been manipulated through the intensity of stimulation; hence the greater intensity and the greater contact force.
Although this review has focused more on peripheral-nerve recording, stimulation, and the methods adopted for encoding, we will briefly address the stimulation issue in the cortical region for restoration of somatosensory feedback.
The field of neuroprosthetics has reached an age of maturity. When similar questions need to be addressed, different strategies are pursued Courtine and Bloch, Elicitation of percept by electrical stimulation was first demonstrated by Wilder Penfield in They showed that somatosensory cortex S1 neurons are organized into separate columns representing different regions of the body Penfield and Boldrey, They later induced tactile sensation by applying electrical stimulation to the somatosensory area of a neurosurgical patient Rasmussen and Penfield, The locations of percept areas on a body are systematically the same as those required on the cortical surface of the brain for stimulation Rasmussen et al.
The neurons that can react to similar types of stimuli have their functional columns in area S1. Studies also have found evidence for S1 regions that encode sensory information for individual digits of a hand along with different types of receptors within those digits Merzenich et al.
Intra-cortical microstimulation ICMS has been applied to primates having the pulse frequency corresponding to the evoked perception of cutaneous flutter Romo et al. Recently, several studies have demonstrated the characteristics of ICMS by varying the parameters of stimulation for sensory feedback and recording of voluntary movements in primates Fitzsimmons et al. Another important breakthrough is the first bidirectional brain-machine interface in which a signal from the motor cortex of a non-human primate was able to control the cursor, while stimulation is applied simultaneously to the brain area S1 to give sensory feedback on the movements, though the primate had to go through a learning process for mapping the afferent interface Rajan et al.
These research advances have highlighted several methods of evoking broader ranges of near-to-natural sensations of touch and proprioception Blank et al. Somatosensation including proprioception is an essential element of natural motor abilities. Without having the sense of proprioception, it is difficult to plan a dynamic limb movement Sainburg et al. Many research groups are trying to use biomimetic approaches to achieve the sense of touch and proprioception integral to everyday behavior Saal and Bensmaia, However, at present, large-scale neuronal activation relating specifically to the elicitation of detailed desired responses is not possible, due to the sensitivity of the brain.
In one study on rodents, the implementation of a biomimetic approach was tested by applying ICMS to the barrel cortex using micro-wires for the generation of artificial tactile percepts to navigate around a virtual target on the screen Venkatraman and Carmena, ; O'Connor et al. With the help of this artificial sensation, the rats were able to find the virtual objects in order to obtain a reward with improved accuracy after training.
Another experiment was conducted, in which the task was to detect the pulses of ICMS over a variable time interval. The rats were able to differentiate between pulses of ICMS and distractor mechanical stimuli to replicate the target stimulus.
ICMS can also be employed to deliver localized tactile percepts with natural features as well as in human-brain-computer interface applications for better control using area S1 Collinger et al. An experiment on the biomimetic method in non-human primates was conducted by delivering ICMS to area S1 in order to convey sensory information on the location of a contact in the hand; it was found that the receptive field of ICMS was highly localized to elicit percepts on the hand or even fingertips Tabot et al.
Various combinations of receptive fields can be used to signal which area of a prosthetic limb is in-contact with an object. The response of a sensor placed on the prosthesis can be used to activate neurons in area S1 with corresponding receptive fields. These approaches and advances can be served as a basis for improved somatosensory feedback in amputees or tetraplegic patients.
In patients, either a limb is missing or its connection with the brain has been lost, though the brain area responsible for tactile stimulation or sensory feedback for the corresponding limb is still intact Flesher et al. These patients often feel a sensation at the residual limb Ramachandran and Hirstein, Task-specific somatosensory feedback has been generated in humans using cortical stimulation via electrocorticographic electrodes Cronin et al.
In the future, patients will be able to learn the meaning behind each combination of stimulation patterns with several types of sensory modalities and, by using that information, operate the prosthesis in more natural and effective ways. Therefore, peripheral interfaces have an edge over cortical interfaces. A substantial progress has been made over the past few decades in the neuroprosthetics technology. There are two types of prosthetics: Myoelectric- and body-powered prosthetics.
Body-powered prosthetics have the advantage over myoelectric-powered ones, because an amputee can touch or feel the interaction with objects through the body harness operating the prosthesis Carey et al. Battery-powered myoelectic prostheses, on the other hand, are based on muscle signals generated from the residual limb of an amputee through surface electrodes.
These signals are transmitted to the actuators of the prosthetic limb to perform programmed movements. To evaluate prosthesis user functionality, several protocols have been developed Chadwell et al.
Pattern recognition control of myoelectric prostheses is also getting popularity due to its better classification of EMG signals user intentions using various methods, i.
Among the most recent achievements is the battery-powered myoelectric prosthesis with 16 degrees of freedom comparable to natural hand motions Cipriani et al. Nonetheless, they provide less sensory feedback than body-powered prostheses. Myoelectric devices have relied only on auditory and visual sensory feedback, while body-powered devices rely on contrast and pressure, and the grip-strength information is transmitted through the hand using a cable and interfacing system.
In some cases, the sensory feedback might not be the same as provided by the user's visual experience, but does provide a sense of control and connection to the device. Fifty percent of users do still prefer body-powered prostheses, possibly for improved feedback or referred sensation Biddiss et al.
The developments in volitional motor control have outpaced the advancements in sensory feedback for better control of prosthetic arms Parri et al. Along with visual feedback for monitoring of motor commands, the prosthesis must have tactile and proprioception feedback.
In any case, choosing prosthetics is always a tradeoff between sensation and function. In comparison to EMG based control, the BCI is also getting popularity for control of prostheses in a non-invasive method. However, the BCI modalities only decode the brain signals using active, passive, and reactive tasks Khan and Hong, ; Hong et al.
The current BCI systems do not have the ability to decode movement intention, instead they use imagination based tasks for control Hong and Nguyen, ; Khan et al. Thus, for an amputee, the invasive nerve implants may be a more effective solution to BCI in comparison with EMG-based control. The technology has been transferred to applications involving direct interfacing with peripheral nerves for intuitive control and sensory feedback.
Different peripheral nerves can be activated through selective electrical stimulation. Hence, tactile or position sensations can be selectively and focally elicited without associated pain, while motor signals to extrafusal fibers are recorded more easily than those to intrafusal.
The advancement in new peripheral interfaces has capitalized on this property to evoke touch perception without eliciting pain. Repeated clinical demonstration with replicated results has been achieved for peripheral interfaces Schiefer et al. Bensmaia and Tyler concluded that many challenges are ahead before selective activation of all receptors. For instance, thermoreceptors for temperature feedback without simultaneous activation of others are required to be achieved yet. Several options are still required to be explored for nerve-machine interfaces Rutten, Upper-extremity amputations are usually associated with significant disabilities.
Daily living activities are either no longer possible or require additional effort and time. This high level of disability serves to emphasize the obligation of the engineering research community to investigate, develop, and achieve innovative technologies for upper-limb prosthetics that can improve the quality of life. In fact, several advances already have been made in the upper-limb prosthetic technologies: However, there remains a lack of truly effective control and sensory-feedback interfaces.
As alluded to in the previous sections, natural sensory feedback in prostheses is highly desirable for amputees. A stable sensory-feedback system can enhance the control of prostheses. In fact, for a clinical use, a stable interface that is capable of selectively evoking several sensory percepts at several locations is indispensable. However, challenges to produce selective stimulation for chronically stable and long-lasting clinical applications for using peripheral nerve interfaces are still remaining.
Although selectivity can be achieved through interfaces that penetrate the nerve for stimulation of one-to-one axons, this is with the cost of increased invasiveness. Additionally, whereas the modern prosthetic technology has achieved greater selectivity through regenerative, intra-fascicular and microelectrode-array-type interfaces, the problems are to test its biocompatibility and long-term stable recruitment in humans.
These properties depend on the proximity of the exposed electrode surface to the target nerve fibers. Stability and selectivity have been achieved for longer periods of time, for instance, more than 3 years using the cuff-type interface Tan et al. Therefore, more experimentation and changes in the design of implanted interfaces can provide selectively long-term graded sensation, tactile perception, and a sense of embodiment to improve the quality of prostheses.
The non-invasive techniques in prostheses have been widely used in myoelectric battery- and body-powered prostheses. Their major advantage is the indirectness in communicating with the amputee or patient , which means that they are non-invasive. However, these types of prostheses also have a severe disadvantage in that they lack wide sensory feedback. Some other modalities, such as the electro-tactile, vibro-tactile, and modality-matched feedback types discussed earlier do not provide appropriate feedback to an amputee and are less intuitive.
In these cases, usability of these prostheses is decreased and the cognitive workload of the patient or amputee often increases. For its achievement, several methodologies have long been studied, based on which the most important goal is to design an interface that is stable and biocompatible and that can selectively acquire signals from the peripheral-nerve.
The most recent study of Wurth et al. Additionally, the same nerve-machine interface is responsible for providing sensory feedback through stimulation that depends upon the area of the prosthetic hand that is in contact with an object.
As noted above, several types of interfaces necessary for the completion of this nerve-machine interface have been designed, tested, and reported in the literature. FINE's more-thanyear stable implantation and sensory percepts in humans has showed its viability, although it obtained only 15—20 sensory percepts without tingling or paresthesia.
For an increased number of sensory percepts at different locations in a hand, multiple electrodes were required as demonstrated by Davis et al. This type of interface configuration, with multiple-electrodes inside the nerve, incurs a greater risk of nerve damage as well as biocompatibility issues.
In this strategy for achieving selectivity at the fascicular or axonal level, the implantation of TIME and LIFE, along with the combination of FINE, will prove beneficial in providing more sensory percepts selectively with an increased number of recording contacts for bidirectional control of prosthesis. Although clinical demonstration of this concept has not yet, to the authors' best knowledge, been achieved, it might prove useful in delivering better results for wide, distinct and stable sensory feedback in the neuroprosthetic technology.
In this paper, we have reviewed the state of the art of peripheral-nerve-machine interfaces for restoration of sensory feedback. The ultimate goal of all these interfaces is to provide a prosthesis with the ability to serve as an actual near-to-natural-limb replacement.
Such prosthesis must be able to perform routine tasks and provide discrete sensation to the amputee. The lack of sensory feedback in myoelectric prostheses or those incorporating indirect methods of stimulation for sensation is a key limitation relative to the achievement of full control.
Many researchers have tried several methods for providing sensory feedback. Among them, the peripheral-nerve-machine interface has become widely popular by providing selective and long-term stable feedback with more or less invasive type interfaces. UG conducted the literature survey and wrote the first draft of the manuscript.
SK participated in revising the manuscript. KH has conceived the idea, corrected the manuscript, and finalized the work.
All the authors have approved the final manuscript. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. National Center for Biotechnology Information , U. Journal List Front Neurorobot v. Published online Oct Author information Article notes Copyright and License information Disclaimer. Bornstein, University of Melbourne, Australia.
Received Apr 22; Accepted Oct The use, distribution or reproduction in other forums is permitted, provided the original author s or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This article has been cited by other articles in PMC. Abstract For those individuals with upper-extremity amputation, a daily normal living activity is no longer possible or it requires additional effort and time.
Introduction According to Ziegler-Graham et al. Open in a separate window. Breaks down of sensory feedback methods used in bionic arm systems. Structure of peripheral nerves The roots of the peripheral nervous system lie within the spinal cord, and the axons spread inside the peripheral nerves to reach the target organs.
Afferent receptors Touch, pressure, proprioception, temperature, and pain fall under the rubric of somatic sensation. Non-invasive methods of sensory feedback Several substitutions for sensory perception have been developed, which do not require implantable interfaces Lundborg and Rosen, ; Visell, ; Khasnobish et al. Electro-tactile stimulation The electro-tactile sensory modality is a method of passing electrical current through the skin of an amputee to elicit perception Yem and Kajimoto, Vibro-tactile stimulation Percept sensation in the residual limb also can be generated using vibro-tactile stimulation elicited by mechanical vibrations of the skin Tanaka et al.
Modality-matched feedback The most recent non-invasive technique used for conveying sensory information is the modality-matched feedback see the mechano-tactile stimulation below: Invasive methods of sensory feedback In the indirect methods noted above, complete selectivity and longevity have not yet been achieved. Neural interface and advantages In the twenty first century, two important developments have transformed the field of neuroprosthetics.
Extra-neural interfaces In general, two broad categories of peripheral-nerve-interfaces are extra-neural and intra-neural. Intra-fascicular interfaces As the name suggests, an intra-neural interface penetrates the protective sheaths. Regenerative peripheral-nerve interface RPNI RPNI consists of an electrode and a residual peripheral nerve, which is neurotized by transacting the nerve and inserting the electrode in between them; it is an internal interface for signal transmission with the external electronics of a prosthetic limb.
Targeted reinnervation TR TR is a nerve-machine interface that has been developed to make prosthetic control and feedback more intuitive. Stimulation site on the chest for sensory feedback. Table 1 Summary of human neural implant studies and modalities for restoration of sensory feedback.
Study Interface configuration No. Repeatedly similar in quality, magnitude, and localized sensory perceptions can be reproduced Tan et al. Also evoked tactile perception with consistent signal-to-noise ratios and percept threshold Oddo et al. Characteristics of stimulation Spikes are responsible for conveying information in the axons of a nerve, and the spike rate or frequency is involved in transmitting sensorimotor commands in a single axon Tyler, Restoration of somatosensory feedback through cortical stimulation Although this review has focused more on peripheral-nerve recording, stimulation, and the methods adopted for encoding, we will briefly address the stimulation issue in the cortical region for restoration of somatosensory feedback.
Current prosthetic technology A substantial progress has been made over the past few decades in the neuroprosthetics technology. Discussion Upper-extremity amputations are usually associated with significant disabilities. The proposed hybrid stimulation scheme for enhanced selectivity and longevity. Conclusion In this paper, we have reviewed the state of the art of peripheral-nerve-machine interfaces for restoration of sensory feedback.
Author contributions UG conducted the literature survey and wrote the first draft of the manuscript. Conflict of interest statement The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Evaluating EMG feature and classifier selection for application to partial-hand prosthesis control. Histologic and physiologic evaluation of electrically stimulated peripheral nerve: Sensory feedback from a prosthetic hand based on air-mediated pressure from the hand to the forearm skin.
Artificial redirection of sensation from prosthetic fingers to the phantom hand map on transradial amputees: Sensory feedback in upper limb prosthetics. Devices 10 , 45— Vibrotactile stimulation to increase and decrease texture roughness. Deep learning with convolutional neural networks applied to electromyography data: Topographical distribution of motor fascicles in the sciatic-tibial nerve of the rat. Muscle Nerve 42 , — Comparative analysis of transverse intrafascicular multichannel, longitudinal intrafascicular and multipolar cuff electrodes for the selective stimulation of nerve fascicles.
Biocompatibility of chronically implanted transverse intrafascicular multichannel electrode TIME in the rat sciatic nerve. Spatial and functional selectivity of peripheral nerve signal recording with the transversal intrafascicular multichannel electrode TIME. Single-trial lie detection using a combined fNIRS-polygraph system. Two-point tactile discrimination ability is influenced by temporal features of stimulation.
Behavioral demonstration of a somatosensory neuroprosthesis. Restoring sensorimotor function through intracortical interfaces: Biological and bionic hands: Consumer design priorities for upper limb prosthetics.
Sensory qualities of the phantom hand map in the residual forearm of amputees. Identifying the role of proprioception in upper-limb prosthesis control: Studies on targeted motion.
Complexes to perform cell viability assays were prepared in the same way but using culture medium instead of water as solvent and without the addition of HCl . As the injured cells are viable and are potentially similar to normal cells, they should be enumerated in the microbiological examination of foods.” The selective . C) PBMC viability after shepherdin treatment. PHA-stimulated PBMCs (harvested at day 3 after PHA stimulation) were incubated with shepherdin[79 – 87].