Functional Electric Stimulation for sensory and motor functions: Progress and Problems

Dimiter Prodanov^1,2, Enrico Marani^1,2, Jan Holsheimer¹

¹ Biomedical Signals & Systems Group, Faculty of Electrical Engineering, Twente University, Enschede, The Netherlands

²Neuroregulation Group, Department of Neurosurgery, Leiden University Medical Center, Leiden, The Netherlands

Address for correspondence: E. Marani, Neuroregulation Group, PO Box 9604, 2300 RC Leiden, The Netherlands, e-mail: e.marani@lumc.nl

Published in Biomedical Reviews, 2003, 14:23-50.

Copyright - pre-acceptance version © 2003 by the authors. However, permission to reprint or republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works, must be obtained from the copyright holder.

To avoid copyright issues, here follows the pre-acceptance version of the manuscript.

Abstract

Functional Electrical Stimulation (FES) tries to restore the lost functions of the nervous system by means of electrical stimulation. FES is recapitulated in this review. An overview is given of the relevant moments in history. The principal notions behind FES are outlined, in which special attention is dedicated to the electrodes for selective stimulation. FES applications encounter specific research problems, which are extensively covered. The major clinical applications are overviewed: pacing of the heart, respiratory pacing of the diaphragm, the rehabilitation and restoration of locomotion and hand functions, and the restoration of bladder and bowel functions. An analysis of the current applications based on the evidence-based clinical approach is performed.

FES for restoration of the lost locomotor functions is a rapidly developing area that is on the verge of broadening its applications. Evidence for this is the increasing number of various neuroprosthetic devices that have been developed recently. Some of them are already beyond the prototype stage. The present FES treatments combined with conventional occupational and physical therapy still remain the most promising approach in rehabilitating Spinal Cord Injury patients and stroke patients. FES in combination with neuromodulation becomes a viable option in the treatment of various urological disorders, however it is still in a rather empirical state of operation.

The presented overview of the FES applications shows that despite the substantial amount of research performed in the field which led to successful development of cardiac and phrenic pacemakers and cochlear prostheses, a number of fundamental scientific problem still remains to be solved before one will see a comparable degree of effectiveness and penetration in common medical practice for the locomotion neuroprostheses and urological FES appliances. The major up-to-date directions of research in clinical and theoretical aspects are summarized.

History of electrical stimulation

Functional Electrical Stimulation (FES) is a part of the broader field of electrotherapy. The term was introduced relatively late in the history of FES by Moe and Post (1) . In general, history of electrical stimulation follows the theoretical and technological breakthroughs in electricity and magnetism and it is based on the developments in Neuroanatomy and Neurophysiology, since electrical stimulation is mainly an application of electro-magnetic theory in the nervous system. Nevertheless, the history of electrical stimulation has its own particular subjects of which the major ones, related to the nervous system, are given in this review.

In the field of Neuroanatomy the application of ethanol to fix brain tissue (Felix Vicq d’Azyr, 1748-1794) opened the microscopic approach for the study of the brain, ending up with the detection of the sub cellular protrusions of the neuron and its supporting cells. One should notice that the detection of the micro-architecture of peripheral nerves (Van Leeuwenhoek, 1719) was remarkably accurate but forgotten for nearly a century. The descriptions of Dutrochet in 1824 (2) and Remak in 1838 (3) once again depicted the tubular structure of the entities in the nerve, together with the detection of myelinated and non-myelinated fibers.

Albrecht Haller (1708-1777) described, the “irritation phenomenon of muscles”: in fact the contraction and relaxation of a muscle. In his notion, it was a ”force” that passed along the nerves that could induce this excitation of the muscles. In addition, Haller made the first concept on sensibility. He found that tissues by themselves lack sensations, but they were noticed and relayed by nerves and their endings.

Luigi Galvani (1737-1798), who developed the so-called “galvanic” element, by means of which nerves could be stimulated, studied the muscle contractions. Later on in 1792, Allesandro Volta (1745-1827) could demonstrate that the galvanic element used electricity. So, the nerves were since considered to be the electric cabling of the body.

Figure 1. Galvanic experiments on human muscles. From G. Aldini, Essai théorique en experimental sur le Galvanisme, avec série d’experiences… Paris, Fournier fils, 1804 Courtesy: Prof. H. Beukers, Leiden University

In the mean time, friction electricity was in use and the first capacitor, “de Leidse fles” (1746), was developed by Muschenbroeck. From 1784 till 1791 the biggest friction-electrical machine was build by John Cuthbertson. Two men had to move two enormous glass discs. Each disc was frictioned by four cushions. The electricity on the glass discs had to be induced into the conductors, a series of Leiden jars. The potential difference reached nearly 330 kV. Van Marum (1750-1837), biologist, physician and botanist used this machine to do electrophysiological research for ten years. This shows that around the beginning of the XIX century systematic application of electricity in research and medicine had already started.

Figure 2. The enormous friction electrical machine together with the Leiden jars built by J. Cuthbertson. Courtesy: Teylers Museum

By the middle of the XIX century with the discovery of static electricity, capacitance and electromagnetic inductance, physiologists could study muscle contractions, nerve conduction and cell excitation.

In the nineteenth century J. Müller (1801-1858) introduced the law of Specific Nerve Energies: “the kind of sensation, following the stimulation of a sensory nerve, depends not on the mode of stimulation but on the nature of the sense organ with which the nerve is linked”. His own disciple Helmholtz (1821-1894) measured the velocity of the nervous impulse and Emil DuBois-Reymond (1818-1896), also a disciple of Müller, proved the existence of the resting membrane potential in the nerve fibers and also proved that each impulse is directly related to a change in potentials.

Sir Charles Bell (1774-1842) and Francois Magendie (1783-1855) established the neuronal basis for the reflex. The dorsal root contains afferent, sensory information, while the ventral root relays the motor function.

The final idea that the central nervous system is the controller of the organism was brought forward by T. H. Huxley (1825-1895) in his Manual of the Anatomy of the Invertebrate Animals (1817) and became generally accepted. This concept became the fundament for the effective application of electrical stimulation. In this period medical applications were numerous but short lived and poorly understood. Electricity was reported to cure rheumatism, neuralgia, insomnia and even cold feet (4) .

In XX century, the major technological advances in electronics and computing (discovery of the diode, triode, transistor and integrated circuits) lead to the development of contemporary FES appliances. The main push in the applications of FES was the successful development and clinical applications of the heart pacemakers.

Literature shows controversy over who was the first inventor of the artificial heart pacemaker (5) . In 1932 Dr. Albert Hyman described an “artificial pacemaker” that he had used with experimental animals to resuscitate the “arrested” heart. About that time, the Australian physician Mark C. Lidwill, along with the physicist Major Edgar Booth, built a portable pacemaking unit. It was demonstrated in 1931, while Hyman developed his device in 1930-31. Lidwill's apparatus had one pole applied to the skin and another into the appropriate cardiac chamber (6) . Later on, the Hyman device was successfully used in patients

The application of cardiac pacemakers to stimulate the human heart dates as far back as 1954. However, it was not until the 60s when pacemakers were adopted on a great scale in clinical practice. Their potential for different applications in the beginning was not recognized both by the medical profession and by the industry. The other reason was the immaturity of the technology and the implantation techniques. Stimulation was carried out by skin electrodes, which left uncomfortable burns when used for a couple of days or more. Later, electrode wires leading through the skin were tried, but infections along the wires were an unsolvable problem (6) . Its solution was the implantation of the entire pacemaker. Dr. Rune Elmqvist designed the world's first implantable pacemaker. The first 10 patients for the implantable pacemaker were operated in 1960 (7) . In the 1970’s, heart pacemakers demonstrated the reliability of implanted electronic devices and showed the benefits in terms of improved quality of life.

In parallel went the attempts to apply stimulation of the diaphragm for respiratory pacing. For some 200 years, electricity has been applied to the phrenic nerves to activate the diaphragm. However, the actual therapeutic opportunity arose in the 1950s, at the same time the heart pacemakers were developed. The first attempts to pace the diaphragm with implanted electrodes were carried out between 1948 and 1950 by Sarnoff and co-workers (8) . One of the most important prerequisites of the routine clinical applications was the introduction of the long-term electrical stimulation by the radio frequency inductive method introduced around the end of the 1950s (9) . In 1966, Dr William Glenn at Yale University performed the first clinical application to patients with hypoventilation of central origin. The ventilation support was made by radio frequency pacing of the diaphragm. Pacing for total ventilatory support in patients with respiratory paralysis was applied in 1971. In the early 1980’s the first commercial phrenic nerve pacers were introduced.

As a result of the successful application of the pacing of the heart, the technological advances achieved during the development of the pacemakers were also successfully applied in other neural prostheses. Restoration of hearing to the profoundly deaf was an early target.

A report by Gersuni and Volokhov in 1936 (10) indicated that hearing could be produced over a normal frequency range, and that it persisted following the surgical removal of the tympanic membrane and ossicles. One of the first recorded attempts to stimulate the auditory nerve was made by Lundberg in 1950, who employed sinusoidal currents during a neurosurgical operation. The patient could only hear noise.

However, a milestone for the development of the auditory prosthesis was the electrical stimulation of the acoustic nerve of a deaf man by an implanted electrode achieved by Djourno and Eyries (11) . They used an implanted induction coil connected with one end to the inner ear electrode and with the other to an indifferent electrode at the temporalis muscle. Another coil, outside the body was held over the implanted coil to inductively transmit the signal from a microphone. The usefulness of the device was very limited since the patient could recognize only few words from the transmitted signal ('papa', 'maman', and 'allo'). This operation, however, opened the doors for further attempts to interface the sense organs.

The early investigators were subjected to severe criticism by both neuroscientists and clinical professionals for the insufficient safety studies and unsatisfactory design of the implants used. The safety of the cochlear implantation improved in the late 1950s with the evolution and developments of microsurgical techniques and stapes surgery. The indications and contraindications for this implantation were elaborated in a broad debate between the clinicians and the promoters of the cochlear prostheses. Questions were also raised about which device should be used and about the safety of the electrical stimulation. The seventies could be considered as a decade of controversy concerning the implant itself and its use. Finally after more than twenty years of on-going experimentation, in 1984, the cochlear implant received FDA approval for use in adults in USA. Nowadays the cochlear implant is a historical example of how a neural prosthesis can revolutionize treatment.

In 1929, Foerster described the effects of electrical stimulation of the human visual cortex (12) . The subjects reported for “perception” of a small point of light during stimulation, which was later called a phosphene This result was reproduced many times in both sighted subjects and blind subjects. The idea that concurrent stimulation of many sites in the brain could produce a single coherent image was postulated by Krieg (13) , based on the retinotopy in the visual cortex.

As a second milestone for the development of neural prostheses can be accounted the implantation of prosthesis in the visual cortex of a blind woman in 1968 by Lewin and Brindley (14) . The prosthesis did not benefit the patient much (she reported only about occasional phosphenes in her visual field), but demonstrated that it is possible to transfer large amount of information in the visual system. Giles Brindley, one of the greatest protagonists in the FES field, attained recognition and acceptance of the applications of neural prostheses in the clinical field.

The results obtained by the Brindley implant and the early cochlear stimulator showed the need to stimulate selectively small groups of neurons or fibers in order to transmit meaningful information into the central nervous system.

Attempts to restore the lost functions of the paralyzed leg muscles were performed for the first time in 1961 by Liberson and colleagues (15) . The system was developed to compensate for the “drop foot” problem in hemiplegic stroke patients. By stimulating the peroneal nerve, the prosthesis triggered ankle dorsiflexion, eversion and inversion. Since 1961, a number of neuroprostheses for restoration of walking, hand and arm functions have been designed and tested with various patients and with different levels of success.

Apart from these well-endorsed success stories, other applications of FES still suffer from major difficulties regarding the clinical acceptance and patient compliance. Two examples are given here:

Long and Masciarelli introduced prosthesis for the hand that could not meet the expectations of the patients and it was later on abandoned.

In the 70s, attempts to treat refractory epilepsy were also carried out by means of stimulation of the cerebellar cortex. Later on, the performed double blind clinical trials did not confirm the claimed effects (16) .

In conclusion: History of FES shows that pacing of the heart emerged at different places and its success stimulated pacing of the diaphragm.

The criticism of the early cochlear prostheses could be surmounted because of tenacious research. However, one might consider these first successful applications as “lucky shots”, based on the simple reactions of the involved organs:

For the cardiac muscle and the diaphragm – simple contraction, and since the tonotopically organized inner ear can easily decode the transmitted signal, the success of the cochlear implant was understandable.

Neuroprostheses for locomotion as well as those for reaching and grasping, on the other hand, require selective stimulation of the fibers in the nerves, which innervate the muscles making the movements. Therefore, in the case of selective nerve stimulation, the research should be directed towards localization of the types of (myelinated) axons and their spatial organization.

FES tries to restore the lost functions of the nervous system by means of electrical stimulation. This can be achieved by either rehabilitation thus improving recovery or by prosthesis (artificial substitution of a dysfunctional part) of the lost function or orthosis (device around organs in the case of impairment) of the diminished function. Example of the latter is the peroneal stimulator for foot-drop and of the former the stimulators of the phrenic nerve. FES is also sometimes referred to as Functional Neuromuscular Stimulation or FNS. This approach is limited only to the sensory and the motor functions. The initiation of reflex activity is, though, not the aim of FES but may be a secondary and an useful collateral effect.

Apart from the clinically recognized “success stories” of electrical stimulation FES for restoration of the lost locomotor functions is a rapidly developing area that is on the verge of broadening its applications. Evidence for this is the increasing number of various neuroprosthetic devices that have been developed recently. Some of them are already beyond the prototype stage. That is why this paper will deal further on primarily with the electrical stimulation of the motor system.

Principal notions in FES

Electrochemical processes at the tissue-electrode interface

Electrical stimulation of biological tissues with metal electrodes requires the flow of ionic charge in the biological tissue. This flow is induced by both capacitive and faradic mechanisms. The capacitive mechanism involves periodic charge and discharge of the electrode double layer. There is no charge transfer across the electrode/electrolyte interface. This is an ideal mechanism for charge injection, but it is limited by the maximal amount of charge that can be transferred before the dielectric breaks down. This is approximately 20 mC/cm². Because the charge required for physiological stimulation exceeds this limit, all electrical stimulation is performed by faradic charge injection. The faradic mechanism involves charge transfer across the electrode/tissue interface and, therefore, electrochemical reduction/oxidation processes. These reactions can be either reversible or irreversible. All irreversible electrochemical reactions are undesirable since they alter the chemical composition of the extracellular fluid, producing cytotoxic products or bringing about large changes in pH. For every material used in the production of electrodes, there is a charge limit for the reversibility of the electrochemical process. This charge limit depends on the particular properties of the material, the shape of the electrode, and its size, and the stimulation waveform. The temporal shape of the stimulation wave is the most important according to Robelle and Rose (1990) (17) .

In order to diminish the oxidation and dissolution of the electrodes generally a noble metal like Pt, Ir or Au is used as material for the anode. Of the non noble metals 316L stainless steel is also used for intramuscular electrodes (18) . At the cathode, however, O₂ molecules are reduced to O₂^-, which in turn react to produce free radicals. The free radicals are known to damage the cell membranes and DNA molecules. During a typical neurostimulation pulse, a substantial amount of the O₂ molecules is reduced to anions (19) .

Part of the free radicals generated near the cathode can be neutralized by oxidation when an anodic pulse follows the cathodic pulse. Morton et al. (1994) (19) concluded that the reduction of oxygen, and thus the generation of free radicals are restricted most when the cathodic pulse has a small width and when an anodic pulse with the same charge immediately follows the cathodic pulse. This principle is known as biphasic charge-balanced stimulation (20) . The continuous stimulation with biphasic, charge-balanced pulses without delay between the pulses are more beneficial with regard to nerve degeneration than those with delays (21) .

The continuous stimulation with biphasic, charge-balanced pulses was shown not to have harmful effects to sciatic nerves (22) .

Voltage-controlled and current-controlled stimulation

Historically the implantable FES systems in clinical applications have generally been voltage-controlled, as for example are cardiac pacemakers. In contrast, the current-controlled devices dominate in experimental set-ups. In the voltage-controlled devices, the output voltage V of the pulse generator is regulated. Generally V is kept constant, thus creating a rectangular voltage pulse, the related current obeying Ohm’s Law. In contrast, in the current-controlled devices the current I is kept constant, whereas V is automatically adapted according to the value of the impedance Z, which may vary over time due to the reaction of the surrounding tissue to the implant. The current needed for stimulation does not change chronically and thus no corrections are needed. In the voltage-controlled systems, however, the voltage needed for stimulation is influenced by the changing value of Z over time and thus the amplitude of stimulation has to be adapted at least in the first two months after implantation when the fibrous encapsulation layer develops.

Since the current consumption determines battery life of an implanted pulse generator, it is important that the stimulation current is determined in the first place. This is, however, not possible when a voltage- controlled pulse generator is used, since I depends on the unknown value of Z. Therefore, the battery life cannot be accurately predicted from V. Nevertheless, it is often erroneously assumed that a reduction of V reflects a reduction of I and vise-versa.

Monopolar, bipolar and tripolar stimulation

Since the cathodic threshold for nerve fibers is 3-7 times lower than the anodic threshold current, cathodic stimulation is by far the most efficient way (23) . In monopolar stimulation, the active electrode (in or near the neuronal target) is therefore the cathode and the distant, indifferent electrode, is an anode. In that situation, the current injected by the cathode is distributed more or less evenly in all directions.

Activation of nerve fibers always happens near a cathode in bi-, tri- and multipolar stimulation. For the optimal stimulation, in bi-, tri- and multipolar stimulation, the position of the cathode is more important then the position of the anode. In the clinical applications of FES phenomena such as cathodic block, anodic excitation and anodic block are not likely to occur (24) , (25) . The threshold current for stimulation of a nerve fiber increases with the distance between the cathode and the fiber and is inversely related to the fiber diameter. The preferential stimulation of large fibers is favored most when short pulses (~ 60 ms) are applied. Smaller fibers can be activated more easily when pulses are wider. The current consumption is minimized when stimulation is given with a “guarded” cathode parallel to the fiber bundle as in peripheral nerve stimulation and Spinal Cord Stimulation (SCS).

Cable properties of the nerve fibers

The theory of the current flow in electric cables was developed for submarine cables by lord Kelvin (1855). It was first applied in studies on excitable cells towards the end of XIX century.

Cable theory opens the possibility to derive equations for current flow in cylindrical nerve and muscle fibers subjected to voltage changes which are small enough for the membrane properties to be linear; or equations that describe finite cables, since in most experiments the length of the preparation is not large enough, compared to the space constant l for the cable to be regarded as infinite.

This means that the behavior of the impulse in an axon can be predicted.

Active properties of the nerve fibers

The nerve fibers conduct information by means of action potentials. The action potentials propagate at a velocity proportional to the fiber diameter.

Table 1. Morphology and electrophysiology of the fibers in Peripheral nervous system

Fiber type	class	MCV¹ (m/s)	Diameter (μm)	Function
Aα	myelinated	70-120	12-20	motor somatic muscle fibre; proprioception; muscle spindle annulospiral; proprioception; Golgi tendon organ
Aβ	myelinated	70-120	5-12	proprioception; muscle spindle flower spray; exteroception; touch & pressure
Aγ	myelinated	30-70	3-6	motor somatic muscle spindle
Aδ	myelinated	12-30	2-5	exteroception; pain temperature (some ) touch
B	thin myelinated	3-14	<3	motor autonomic preganglionic
C	unmyelinated	.5-2.5	0.4-1.2	exteroception; pain reflex responses
C	unmyelinated	.5-2.5	0.3-1.3	motor autonomic postganglionic sympathetic

¹Mean Conduction Velocity

Moreover, the threshold of excitations also depends of the fiber type.

In the myelinated fibers, the conduction is saltatory, with the active involvement of the nodes of Ranvier. For the first time the action potential of the non-myelinated fibers were mathematically described by Hodgkin and Huxley in 1952 (26) .

The electrical behavior of the myelinated nerve fiber can be represented by a simple cable network. McNeal was the first to introduce this model (27) in order to calculate how a stimulation-induced extracellular field affects nodal transmembrane voltages. According to this model, the node of Ranvier that is closest to the cathode will be excited first when the stimulus current is sufficiently high. The results of this theoretical approach are in accordance with the general observation that the threshold stimulus of nerve fiber excitation is smallest near the cathode and rises with the increase of the distance (23) . As a result, cathodic stimulation gives rise to a depolarization of several nodes, whereas a larger number on both sides of the cathode are hyperpolarized to a lesser extent, thus creating a virtual anode.

Figure 3. The McNeal cable model for the axon. Each node of Ranivier is represented by an equivalent electric scheme. Vi – intramembrane potential, Ve – extamembrane potential, Ra –intraaxonal resistance, R – variable membrane resistance, C- membrane capacity, V – resting membrane potential

In anodic stimulation the opposite effects occur. The virtual anode and virtual cathode effects underlay phenomena such as cathodic block and anodic excitation. Anodic excitation is possible when the anodic current is so large that the virtual cathodic depolarization on either side of the hyperpolarized membrane generates action potential. In cathodic block, the opposite is true – if the cathodic current is large enough the induced virtual anodic hyperpolarizations can block the propagating action potential.

The McNeal model is the basis of all the theoretical research performed in the area of FES.

Stimulation Electrodes

Spatially selective stimulation of the nerve fibers within a given nerve bundle is important, because different muscles are innervated by fibers possessing a specific localization within a nerve bundle, and moreover often a mixture of efferent and afferent fibers is present in the nerve. Such mixed populations may lead to undesirable sensations during stimulation. In the case of several nerve fibers innervating different muscle fibers of one muscle, cyclic stimulation can be performed which guarantees an overall stimulation frequency, while stimulation frequency and thus fatigue of a single muscle fiber can be kept low (28) .

Many different designs of electrodes exist. They employ different materials and geometries of the contacts. Most of them are used only in animal experiments, and only few are employed in clinical FES applications as well.

Several types of stimulation electrodes for FES have been developed. A simple subdivision is

Surface (skin) electrodes

Intramuscular and epimysial electrodes

Extraneural electrodes (cuff or helical electrode around a peripheral nerve)

Intraneural wire electrodes

An extensive account on the cuff, intrafascicular and sieve electrodes designs and fabrication technology can be found in (29) and (30) .

Surface electrodes

Transcutaneous stimulation is performed with self-adhesive or non-adhesive electrodes that are placed on the subject’s skin, above the peripheral nerve (respectively the motor fibers). The metal electrode is connected to the skin with a saline bridge.

Criteria for surface electrodes are low impedance with even current spread, flexibility, ease of application and removal and lack of skin irritation (31) . The application is simple. Damage of the skin only can occur by the diffusion of soluble electrochemical products. The large distances to the stimulation targets and the insulation of the skin and the fat tissue demand high stimulation strengths, which in turn results in low selectivity. Complications are rare and include burns, skin irritation, erythema and local pain.

Intramuscular and epimysial electrodes

Muscle electrodes have their stimulating surfaces onto or inside a striated muscle. In the former case, they are sutured to the epimysium of the muscle and, therefore, are called epimysial. If terminal motor branches are present, the electrodes exert their action by stimulating the nerve endings. Therefore, it is correct to consider them also as nerve electrodes. In comparison with the transcutaneal electrodes they produce contractions with lower currents and with greater selectively. For these reasons, muscle-based electrodes are preferable for situations that require independent control of several isolated muscles.

Depending on their intended application, intramuscular electrodes can be introduced either percutaneously (see for example (32) ) or in an open surgical procedure (33) . However, intramuscular electrodes can also activate neural structures near their stimulating tips other than those intended, thus recruiting muscles in addition to the ones targeted (34) . Nevertheless, intramuscular electrodes are considered safe and effective means to produce strong and isolated contractions of single muscles.

Complications include infections and wire breakdowns. In an effort to avoid the use of percuataneous leads for muscle electrodes, Loeb (2001) (35) and Cameron et al. (1997) (36) developed implantable muscle stimulators that can be controlled telemetrically, and are inductively powered from external magnetic devices in order to overcome infections. According to the promoters of such an approach, this would ultimately alleviate the burdens involved with maintenance of the awkward external electronic equipment.

Cuff electrodes

Cuff electrodes circumscribe the nerve. They were introduced in the 1970s. The first cuff electrodes had a very simple cylindrical rigid design. Modern cuff electrodes use flexible materials and adaptable geometries.

They are implanted by a simple surgical procedure, and their size can be adapted to the diameter of the nerve of interest. If needed, they can be surgically fixed to adjacent tissues.

Various designs have been developed and applied such as the helix-shaped electrode (37) , the spiral-cuff (38) , and the so-called "half-cuff" electrode of (39) . Another particular design is the arrangement of flexible interdigitating sub-units with microelectrodes along a backbone-like carrier (40) .

The main push for their continuing development was the inadequate spatial and fiber selectivity they provided at first. The first cuff electrodes could stimulate only the most superficial fibers of the nerves (29) .

With a sufficient number of electrodes within the cuff, high spatial selectivity of stimulation can be achieved by combinations of longitudinal and transversal currents produced by appropriately switched electrodes (41) . Longitudinally aligned tripolar dot electrodes on the surface of a nerve trunk restrict excitation to superficial nerve trunk regions more successfully than will monopolar dot electrodes do (42) . Selectivity was dependent on the relative location of the electrode contacts and the nerve fascicles, as well as the size and relative spacing of neighboring fascicles of a 12 contact spiral nerve cuff electrode (43) . On the basis of computer modeling results Deurloo et al. (2003) (44) concluded that transverse steering provides more selective stimulation than longitudinal steering. However, the problem of different fiber diameters, which leads to different activation thresholds of the fibers, is not solved by these methods. Goodall et al. (1996) found that large fibers were activated before smaller with a cuff electrode containing 12 electrodes arranged in four longitudinal tripoles, irrespective of the fiber position (45) . Transverse current from an anode positioned opposite to the stimulating cathode was found to improve spatial selectivity, and position selectivity was enhanced when the ratio of transverse current to longitudinal current was increased.

The use of silicone and helical designs in cuff electrodes has improved surgical access and reduced nerve damage (46) , (47) and (41) .

Figure 4. A schematic view of a cuff electrode. Courtesy: K Deurloo, from (96) p. 66

Figure 5. (a) Scheme of the University of Twente 128-electrode 3-D glass-silicon array (UT-128 array), mounted on a CMOS , mixed mode, processing chip with dimensions 4x4 mm. Needle length is 600, 425 or 250 μm, width at tip is 15 μm, needle spacing is 120 μm. (b) Details of the dimensions and materials used for the UT-128 array. (c) A ‘sea’ of sawn and etched silicon needles of three different lengths, embedded in a glass matrix. Courtesy: W. Rutten, from (30).

Intraneural electrodes

In an attempt to overcome the selectivity problems with stimulation cuff electrodes intraneural electrodes were developed. They are another type of electrodes directly contacting nerve fibers within a given nerve bundle.

Intraneural wire electrodes

The first penetrating electrodes were simply thin metal wires or needles, which were inserted into the nervous tissue. They are used for both recording and stimulation. An interesting, non-silicon approach for intrafascicular stimulation is the use of tethered platinum microwires. Nannini and Horch (1991) (48) tested the performance of Pt-Ir intrafasicular electrodes (25 mm diameter) implanted in nerves innervating the gastrocnemius and soleus muscles. The electrodes were originally developed for recording purposes. They concluded that intrafascicular recording electrodes are also a potentially suitable technology for FES. Yoshida and Horch (1993) (49) used dual intrafascicular electrodes to study activation of nerve fibers by pairs of Pt-Ir wire electrodes implanted within single fascicles of the nerve innervating the gastrocnemius muscle in cats, in order to determine whether intrafascicular electrodes can activate nerve fibers in different fascicles independently of each other and if they can be used to activate separate subsets of axonal populations within a single fascicle.

Intraneural multi-electrode arrays (MEAs)

A simple classification consist of

a) Linear multielectrodes:

Wedge-shaped microprobes carry a line of electrode sites for recording and stimulation, the so-called "one-dimensional" (1D-) arrays (50) .

b) Two- and three-dimensional multielectrodes

The 2D-array consists of electrodes with tips ending in the same plane. The tips are configured either as a bundle of wires or galvanically grown needles. The substrate carrying the electrodes is either needle- or wedge-shaped to allow penetration of the nervous tissue which makes recording from and stimulation of axons possible not only on the surface but also in a well-defined depth within the tissue, e.g. within the fiber bundle. When the array is slanted or the needles have otherwise variable lengths the electrode is named a 3D-array. Implantation of any such a device is always associated with some damage of the nervous tissue, cell death of neurons and disruption of axons. Moreover, stiffness of many models may also lead to mechanical damage of nervous tissue. Thus, the efforts are directed to miniaturize the penetrating parts of the implant and to use materials that are more flexible.

Rutten et al. produced a 3D needle array with 128 recording sites on one electrode placed on the tip of a needle (51) . The needles are made of silicon and are embedded into a glass substrate. They vary in height from 250 to 600 um and have a distance of 120 um, with a tip size of 15×15 um. The array is intended to serve as an interface between a stimulator and a peripheral nerve. The different length of the needles allows stimulation in the volume of the nervous tissue. The dimensions of the array were based on calculations for optimal stimulation, based on random uniform distribution of the nodes of Ranvier in the peripheral myelinated axons.

For signal recording and stimulation over a larger area, e.g. areas of the cortex, two-dimensional MEAs are necessary. An example of such a 2D array is the Utah Electrode Array initially developed by Nordhausen and colleagues (52) . It is designed for implantation in the human visual cortex and is intended as prototype in development of visual prostheses (53) . It provides a multichannel interface to the visual cortex using a large number of 1.5mm long electrodes (typically 100 in a 10 x 10 square grid) projecting out from a very thin (200 um) substrate and are separated from each other by 400 um.

With the technological advance in the field of constructing flexible polymeric substrates, so-called "flexible nerve plates" came into development (54) . They consist of a flexible substrate, which carries a Micro Electrode Array and it is possible to add other elements of a microelectronic circuitry.

In general the silicon microtechnology based fabrication leads to improvement of recording structures, which made it possible to create not only planar two-dimensional Micro Electrode Arrays but also intraneural electrode structures for in vivo measurements. Alike the cuff electrodes, geometry and arrangement of the intraneural electrode arrays are currently optimised to fit with geometrical nerve fiber characteristics (fiber position in a bundle, diameter of the fibres and their orientation). While Smit (1996) argues for intrafascicular stimulation, the proposed approach could be employed only in acute experiments, thus limiting the use of the obtained knowledge for clinical applications (55) .

In theory intraneural electrodes should give better spatial selectivity than the cuff electrodes (56, 49), but the distribution of the nodes of Ranvier in the peripheral nerves makes the stimulation a probabilistic process. The implanted intraneural electrode may simply miss the target fiber population when the topography of the nerve fibers is unknown.

Also since during implantation these intraneural electrodes must penetrate the relatively tough epineurium and the perineurium of the nerve, they can cause short- and long-term nerve damage during the electrode insertion procedure (57, 58, 59, 60 ). Nevertheless, histological results suggest that intraneural electrodes can be used chronically in peripheral nerves (61, 62, 63).

The main prerequisite for successful application of the modern sophisticated electrode designs and stimulation protocols is detailed knowledge of the topography of the nerve fibers within the nerve bundle and their function. Moreover, a really reliable fixation of the cuff must be achieved because otherwise the cuff may rotate around the nerve or shift along the nerve, both leading to a serious loss of selectivity.

In conclusion: The implantable electrodes are surgically set in: around, within or adjacent to a nerve trunk or root. By means of transcutaneous electrodes it is difficult or impossible to selectively activate individual muscles, especially deep muscles such as hip flexors (64, 65). Large currents may be required to drive sufficient charge through the skin and intervening tissues between the electrode and the peripheral nerve. In many cases, cutaneous pain receptors are excited, and patients with preserved or heightened sensation may find it difficult to tolerate surface stimulation at the levels required to produce a functional motor response.

Advantages of the implantable electrodes over the superficial, epimysial and intramuscular include:

The current needed to excite an action potential in the target fibers is smaller in at least an order of magnitude over the current required for the latter designs.
Muscle length and limb position are expected to have substantially less effect on the recruitment characteristics in the case of an implantable electrode.

Cuff-electrodes are easy to implant and are generally less invasive to the nerve than intraneural electrodes are.

Despite all new developments in electrodes, the robustness of stimulation is hampered by the structure of the nerve:

The typical fasciculation

An ideal peripheral nerve electrode array would have the neural selectivity of the intraneural electrode and the ease of implantation, stability, and biocompatibility of the extraneural electrode.

Research Problems

The major issues in the development of a neural prosthesis from the point of view of the neuroscientist are:

Biocompatibility of the implantable devices: general considerations

Implantable devices should be able to survive in the rather aggressive internal milieu of the body without causing significant tissue reaction. The major prerequisite for the application of implants is that the organism accepts the implant, i.e. that the implant is biocompatible. This also holds for objects that come in close contact with the skin or mucose. An accepted definition of biocompatibility is "the ability of a material to perform with an appropriate host response in a specific application" (66) . The selection and evaluation of materials and devices intended for use in humans requires a structured program of assessment to establish biocompatibility and safety.

Current regulations, whether in accordance with the U.S. Food and Drug Administration (FDA) (ISO 10993-1/EN 30993 standard, since 1995), the International Organization for Standardization (ISO), or E.U. regulation bodies (The EU council directive - 93/42/EEC, since 1993), as part of the regulatory clearance process require conduction of adequate safety testing of the finished devices through pre-clinical and clinical phases (67) . An extensive account on the biocompatibility may be found in the standard ISO 10993-1/EN 30993.

In brief, the notion of biocompatibility encloses non-toxicity, non-immunogenity, non- carcinogenicity, non-mutagenity and haemocompatibility. An implant can be considered biocompatible if it gives negative results on the following tests:

Irritation Intracutaneous

Acute Systemic Toxicity

In other words, it includes the whole behaviour of the implant in its biological environment.

According to Heiduschka and Thanos (1998), the "biosafety" and the "biofunctionality" also have to be considered (29) . Biosafety means that the implant does not harm its host in any way, and biofunctionality means that the implant acts in the body as it was intended. In addition, "biostability" is important which means that the implant must not be susceptible to attack of biological fluids, proteases, macrophages or any metabolic substances. For example, implants may be subject to continuous attack by hydrolytic enzymes (68) or free radicals produced by monocytes and/or cell lysis. Stability of implanted material is important not only for a stable function, but also because degradation products may be harmful to the host organism. Overview about biological reactions to implanted materials can be found in Ratner et al. (1996) (69) .

Chemical biocompatibility

Chemically appropriate implant materials are designed to be as inert as possible. If chemical reactions are to be expected they should be minimal and all resulting products should be inert. Candidate materials for neuroprothesis pass especially rigorous testing since they must remain inert not only passively but also when subjected to electrical stimulation. Typical materials for nerve electrode arrays include platinum-iridium alloys or stainless steel for the conducting parts and epoxy resins, polytetrafluoroethylene (PTFE, Teflon^®), silicone rubbers and polyimide for insulators (70, 29) . These polymers are biocompatible, electrically insulating and stable.

The amount of platinum ions released into the surrounding tissue may be neglected even after long-term stimulation. During the last years, iridium has been of increasing importance because a stable oxide film can be formed on the surface of iridium electrodes. This oxide film has a big charge delivery capacity and is, for this reason, well suited for stimulating electrodes. Platinum and iridium are established materials in microelectronics, and carbon can be deposited onto microelectronic structures. Glassy carbon or carbon fibers are also used as electrode materials, and they are biocompatible and stable, though they have a higher roughness than metals.

The bulk properties of polymers as backbones for electrodes can be modified to a certain degree and also surface modification procedures are performed in order to improve biocompatibility. However, certain surface modifications like autologous protein coatings also enhance the adhesion of pathogenic bacteria to the implant surface (71, 72). For a review on the chemical modifications employed in the production for neural prostheses, see (29) .

Mechanical biocompatibility

The mechanical biocompatibility of an implant depends on the basic mechanical properties of the tissue and the implant. Ideally, the implantable device should have mechanical properties similar to the tissue in which it is implanted. This includes flexibility, strength and durability. It is also important to consider the tissue motion within the implantation site. For example, flexibility is important in the peroneal foot-drop stimulator and not in the devices for spinal cord stimulation and anterior lumbo-sacral root stimulator.

For any implant, the allowable size must be defined relative to its function. It is not necessarily true that increased function is associated with larger size. The geometrical characteristics of an implant are of great importance too. For example, sharp edges and blunt corners should be avoided.

The primary goal for a cuff electrode design is to minimize the size, however any constriction injury in the implanted site should be avoided. Such designs have been associated with incidence of morphological changes in neural tissue like axonal degeneration, demyelinization and fibrosis (37) . Thus, the recommendation had been to implant cuffs with an internal diameter equal to 150% of the nerve diameter (73) . This recommendation was criticized as to be actually more likely to create neuronal trauma (37) . It also limits the degree of selectivity that is obtained during extraneural stimulation (41, 31) .

Implantation and the Nerve damage

The key factor determining the damage of electrode implantation is the disruption of the microcirculation of the nerve, which results in edema. This holds especially for intrafascicular electrodes. Another factor is the damage to the endoneurium and the disruption of the homeostasis, maintained by the blood-nerve barrier. The mechanisms of the damage during implantation are still debatable. According to Agnew and McCreery (1990) on the first place stands direct mechanical interaction between the electrode and the nerve (70) . Other mechanisms that may play a role during implantation are:

Surgical trauma to either the neural microvasculature or the nerve itself.
Pressure caused by post-surgical edema, seroma formation, or excessive fibrous encapsulation of the implant.
Reduced mobilization of the nerve caused by excessive scar tissue formation, which could fuse the nerve to the surrounding tissues.
Undue tension in the electrode leads.
The transmission of forces from muscles to the electrode array and the nerve.
Obstruction of the microstimulation

For epineural and intraneural designs, though, the most important factors are the implant procedure itself and transmission of tension via the leads.

Tissue reaction to implantation

Implantation usually causes foreign body reaction involving the natural inflammatory defensive response of the body. The acute response starts with protein adsorption onto the surface of the implant. Therefore, the surface properties like roughness and surface chemical composition, which depend on the production process, are important for the grade of the subsequent response (74) . The late response produces the typical cellular and humoral reaction including tissue edema, cytokines secretion, neutrophil, lymphocyte and macrophage migration and adhesion. This is a complex process mediated by the secreted cytokines.

Chronically a fibrotic capsule around the foreign body is formed, consisting of collagen fibers and fibroblasts. The thickness and the structure of the capsule is generally a measure of the severity of the response. The severity of the reactions can be determined by the amount of the giant Langerhans cells and macrophages.

The electrical properties of this fibrotic capsule were studied by Grill and Mortimer (1994) (75). They found that resistivity of the encapsulating tissue had a frequency dependency between 10 Hz and 1kHz and decreased from 454 ± 123 to 193 ± 98 Ohm.cm and was frequency -independent between 1 kHz and 100 kHz with a mean value of 195 ± 88 Ohm.cm. Moreover, the fibrotic capsule may lead to displacement of the electrode positions and changes in the tissue impedance (75) .

The results from early experiments (37, 76) , indicated that nerves implanted with snuggly fitting spiral cuffs showed signs of sustained trauma. The typical pattern was a crescent-shaped region containing thinly myelinated axons, proliferation of subperineurial connective tissue, and a reduced axon density. In a more extensive study of the tissue reaction to the spiral cuff, 4 of 44 nerves exhibited peripheral crescent-shaped areas which contained thinly myelinated fibers with an apparent reduction in axon density, but there was no correlation between the cuff-to-nerve diameter ratio and the presence of morphological abnormalities. Similar results have been found in cuff electrode studies conducted by researchers at the Huntington Medical Research Institutes (77) .

Recently Grill and Mortimer (2000) (78) reported focal areas of abnormal neural morphology including perineurial thickening, endoneurial fibrosis, thinly myelinated axons, and focal reduction in the density of myelinated axons in chronic implantation experiments with cuff electrodes.

Intrafascicular electrodes can also produce morphological changes in neural tissue (63) . Chronic implantations of intraneural coiled-wire electrodes showed endoneurial fibrosis and edema, loss of nerve fibers with the large myelinated fibers being the most susceptible, and variable shifts in the excitation threshold. However, there were no changes in conduction velocities indicating that no significant damage occurred due to implantation (61) . The bulbous enlargement formed at the point where the electrode penetrated the perineurium was associated with focal nerve fiber compression, demyelination, edema, and fibrosis (61, 79) . Electrodes sutured to the epineurium have also led to neural damage including endoneurial edema, endoneurial fibrosis, loss of axons, and reduction of myelin (80, 81) . In general, intraneural electrodes are associated with greater risk of trauma than the cuff-electrodes (82) .

Nervous tissue reaction to electrical stimulation

Brief stimulation of the peripheral (83) or the cranial nerves results in increased expression of Immediate Early Genes (IEG) in the related neuronal cell bodies (for example c-Fos (84) ). The family of the Immediate Early Genes consists of approx. 30 genes, with related proteins acting as transcription factors (85) . Members of the IEG family in general are activated shortly after cell stimulation and without the requirement for de novo protein synthesis.

Brief unilateral electrical stimulation of the cochlear nerve (120–250 mA, 5 Hz, 30 min) in anaesthetized rats with a biphasic current resulted in increased expression of c-Fos in the ipsilateral ventral and bilateral dorsal cochlear nucleus (84) . Intracochlear electrical stimulation with a cochlear implant in rats lead to changes in the phosphorylation state of the cAMP response element binding protein (CREB) and the expression of the IEG family members c-fos and egr-1 in a tonotopically precise pattern in the central auditory neurons (86) . These neurons resided nearly in all auditory brainstem nuclei. Moreover, effects of electrical stimulation were identified in the medial vestibular nucleus and the lateral parabrachial nucleus. Regionally, CREB was dephosphorylated, wherever immediate-early gene expression went up. These massive stimulation-dependent modulations of transcription factors in the ascending auditory system indicate ongoing plastic changes as a consequence of the stimulation of the inner ear.

In the spinal cord, Molander et al. (1992) (87) showed that in chronically axotomized nerves, c-fos is expressed after electrical stimulation of the fibers. Stimulation of the normal sciatic nerve at C-fiber intensity resulted in c-fos protein-positive cells within the sciatic projection territory in the ipsilateral dorsal horn (lamina I and outer lamina II) and stimulation of A/B fibers had little effect. The expression was delimited to the projection areas of the sensory A/B-fibers (ipsilateral laminae II, III and IV and in the gracile nucleus) in young animals. This suggests that the excitability of these neurons is increased by nerve injury.

The role of the IEGs is implied in the structural plasticity of the nervous systems (88, 89). One may speculate that the continuous improvement of the responses of the individuals after early rehabilitation, in contrast with late rehabilitation, is due to FES being able to guide the plastic phenomena occurring after injury in a direction towards restoration of function. This line of thought is supported also in the review on animal experimentation (90) showing accumulation of physiological and behavioral data that adaptive (plastic) processes also occur within spinal circuits. The potential ability of the spinal cord to “learn” has obvious implications for altering and improving locomotor function after injury.

Recent studies in man indicate that a significant population of stroke and Spinal Cord Injury (SCI) patients could also benefit from FES rehabilitation (91, 92). In particular, it was found that stroke patients as well as incomplete SCI patients subjected to intensive FES treatment were able to recover grasping or walking function faster and better compared to patients who did not participate in the FES treatment post injury.

In the peripheral nervous system, prolonged high-frequency (f> 20-50 Hz) electrical stimulation of a peripheral nerve induces a typical type of neural injury, called early axonal degeneration (EAD), with a characteristic “salt-and-pepper” like mixture of “normal” and damaged fibers. The primary damage is observed as collapse of the myelin into the axoplasm and is restricted primarily to the large caliber axons (Aa, Ab type). It is followed by degradation and phagocytosis of the axon. (77, 93, 82) . The neuronal injury originating from electrical stimulation has two mechanisms of occurrence:

a) Direct injury resulting from the electrochemical process near the electrodes. Brummmer and Turner (1972) (94) have shown that the rate of production of compounds by electrochemical reactions and the type of compounds produced are directly related to the charge density (charge transferred per unit area of the electrode surface). The charge density near the surface of an electrode determines the extent of depolarization and hyperpolarization induced in the neurons close to the electrodes. If such depolarization is abundant and going on for a long time the second mechanism of injury comes into play.

b) The excitotoxic neuronal injury, which is caused by the excitatory neurotransmitter glutamate through its NMDA receptors. This view is supported also by the finding that MK-801, a potent NMDA receptor antagonist, is a neuroprotective factor during prolonged electrical stimulation (95). The mechanism of the induced neuronal cell death is necrotic. Such damage occurs when the stimulus charge density or the charge per phase is large.

Selectivity concepts

In the paradigm of the implantable devices, selective stimulation of individual components of multifascicular nerves would allow control of several muscles with less hardware (43) . Different aspects of selectivity may be distinguished depending on the structures being stimulated. Here we propose modification of the classification of Smit (55) .

Muscle selectivity is the possibility to activate any specific muscle by peripheral nerve stimulation. This implies stimulation of only a particular part of a (peripheral) nerve i.e. spatial selectivity. In other words, it implies control at the level of the individual muscle
Spatial selectivity is the possibility to stimulate a particular region inside a nervous structure.
The peripheral and the spinal nerves generally consist of several morphologically distinct sub-organs, called fascicles. Therefore, a suitable definition of fascicle selectivity is the stimulation of only one fascicle in a multifascicular peripheral nerve without spread of the activation to other fascicles. It is a special case of spatial selectivity. Fiber selectivity is another special case of the spatial selectivity. It can be defined as activation of single nerve fibers.
The nerves consist of axons, which in general have different sizes. Usually the fiber size distribution is multimodal. Size selectivity can be defined as activation of fibers of a particular size group.

Surface stimulation vs. Implantation

An important advantage of the surface FES systems is that they do not require surgical intervention, with its inherent risks. They also can be removed at any time if contraindications arise. Another advantage is that the transcutaneous electrode can be placed directly above the target muscle, and thus muscle-selective stimulation can be achieved (96) . A typical example where surface stimulation can be used is restoration of hand function. Surface FES can be applied at a very early stage of the rehabilitation, during the recovery and reorganization period of the central and peripheral nervous systems (plasticity), allowing early benefit for the patient. FES training during recovery may help a subject to restore a function to the point that he/she no longer needs a neuroprosthesis.

Popovic et al. advocate an early start of FES as part of the rehabilitation in stroke patients. This benefits the use of surface FES systems since the objective of the treatment is to help patients to relearn the grasping task rather than to provide them with a permanent assisting system (91) . In SCI subjects there are two possibilities: implantation and surface stimulation. For the implantation, the patient should have reached stable neurological status.Surface FES is also preferred in treatment of spasticity in order to strengthen the antagonistic muscles and in recovery of simple movements, which requires stimulation by few electrodes with easy positioning (4) .

A disadvantage is that only a small part of a muscle can be stimulated, causing a rapid development of muscle fatigue. Furthermore, the reproducibility of muscle force is insufficient due to changes in the muscle geometry during contraction. Another issue is the displacement of the electrodes, which may occur during use of the FES device (96) .

The advantages of the implantable FES systems are that the stimulation does not depend on the geometry of the muscle and a single electrode can stimulate several muscles with low energy consumption. Therefore, implantable stimulation is preferred in the following cases (4) :

When deep muscles that are difficult to stimulate by surface electrodes have to be recruited

For complex movements which require a large number of electrodes in a limited space

To avoid pain and burns, caused by stimulation of the pain receptors and the skin.

The disadvantages are that the existing implantable FES systems still do not provide enough selectivity of stimulation, the surgical trauma associated with the implantation and some unresolved issues in the geometry of the electrodes (97) .

In conclusion: Implantation of a FES device induces acute, late and chronic inflammatory response. The least effect is caused by the chemical reaction due to the inert device materials used. The biggest damage comes from mechanical stress or is due to surgery.

Contemporary Clinical applications of FES

The cochlear implant

The cochlear implant relies on the assumption that there are enough auditory nerve fibers left for stimulation near the electrodes. Although the first fully implantable cochlear prostheses preceded the fully implantable heart pacemakers, in the 1970’s, heart pacemakers achieved high reliability and improved quality of life of the patients. In the USA, FDA nevertheless approved the first cochlear implant device for clinical use not earlier than 1984. Over the past 20 years of clinical experience more than 20.000 people worldwide have received cochlear implants.

Cochlear implantation has a profound impact on hearing and speech perception in postlingually deafened adults. Most individuals demonstrate significantly enhanced speech reading capabilities, attaining scores of 90 -100% correct on everyday sentence materials (98) . Moreover, according to the NIH Consensus Statement the cochlear implant is the first, and still the only, neural prosthesis that is aiding a significant portion of a disabled population. In recapitulation - while the cochlear implant has reached wide acceptance and maturity as device, most of the other neuroprosthetic applications are still in childhood.

Drop-foot stimulator

The first neuroprosthesis for walking was developed in 1961 by Liberson and colleagues (15) . This system was developed to compensate for the “drop-foot” problem. The drop-foot is a pathological condition, caused by diminished ability to use the muscles that lift the foot. It may be caused by stroke, cerebral palsy, multiple sclerosis, or neurological trauma. By stimulating the peroneal nerve, the prosthesis elicited ankle dorsiflexion, eversion and/or inversion thus allowing the subject to make a step with the disabled leg. Since then a multitude of devices has been developed, mostly in former Yugoslavia. The Fepa system was proposed by Vodovnik and colleagues (1978) (99) . Currently there are several commercial systems all based on surface stimulation - MikroFES (Jozef Stefan Institute of Science, Slovenia), Odstock 2 (Salisbury District Hospital, UK) (100) and WalkAide (Neuromotion, Canada) (101) . The drop-foot stimulators are commonly controlled by a foot switch. A recent overview on the different FES systems can be found in Popovic et al. (2001a) (97) .

So far, most systems are external with a surface electrode over the peroneal nerve just below the head of the fibula.

More recently, radio-frequency transmitter enabled implants have been developed: for example, the new implantable two-channel system (Finetech, UK) (102) with subepineural electrodes in the n. peroneus profundus and n. peroneus superficalis. It has been tested in 10 patients. This dual channel implant will soon be registered with CE-mark. It has the ability to stimulate independently the deep and superficial branch of the peroneal nerve, thus allowing the correction for excessive eversion and inversion.

Odstock 2 is a surface stimulation system that has been used mostly in the clinical environment. It is based on the stimulator proposed by Liberson. The device can be used as an assisting aid or as a training device to strengthen muscles and improve voluntary control. Additionally, the device has a role in physiotherapy in gait re-education allowing isolated components of the gait cycle to be practiced under therapist control. (103, 104) The Odstock 2 was perceived by the users to be of considerable benefit. A comprehensive clinical follow-up service is essential to achieve the maximum continuing benefit from FES-based orthoses. A recent review on the subject can be found in Lyons et al. (2002) (105) and (97). Odstock 2 and the MikroFES, have been fitted to more than 500 subjects. Thus far, only the WalkAid has been FDA approved.

Figure 6. The peroneal stimulator (Finetech, UK) (see text) Courtesy: J. Holsheimer, G. Bulstra and H.E. van der Aa

Restoration of Lower limb Function

In 1963, Kantrowitz reported the first application of FES to a T-3 paraplegic patient who achieved standing by surface stimulation of the gluteus and the quadriceps for brief periods. The first systems were external devices enabling walking and were developed by Bajd and Kralj in the 1970s through 1980s in Lublijana, Slovenia (106) . The initial devices only used two surface stimulation channels for each leg to produce standing and walking. The stimulation was delivered to both knee extensors. Devices with 4 channels have been proposed later (107) . Systems used by patients included stationary bikes for exercise, transfer systems to enable the patient to move between a wheelchair and a bed, transfer systems between standing and sitting, and walking aids (106) . The targeted patient group was paraplegics after SCI. Patients used a rolling walker frame or crutches for support. About 50 patients successfully used the system (108) . One of the first implantable multichannel stimulators was developed by Brindley with electrodes applied on the femoral and the gluteal nerves (109) . An overview of the contemporary devices can be found in (97) . In brief they are Parastep, LARSI (110) ; FESmate, HAS, RGO, Praxis24 and the implanted FES system proposed in (111) .

The only commercialized system, developed initially by Graupe and colleagues (112, 113) , is Parastep (Sigmedics Inc., USA). It is based on the earlier work of Kralj (106) . The target patient group consists of the complete or almost complete T4-T12 SCI patients. The system uses surface electrodes and no orthotic aids. The Parastep system was applied to more than 400 subjects and was the first FDA-approved FES system.

Although FES of locomotion gains clinical acceptance, it is still in its experimental phase of development. At this stage, FES of locomotion is so far incapable to deliver a complete treatment for the target groups of patients. Only in several cases, it may provide pathogenetic treatment. The main impediment of its development is the insufficient selectivity of stimulation and the small number of applications in patients, which hinders assessment of any beneficial clinical outcome.

Hand disabilities – reaching and grasping

In tetraplegic and stroke patients, the most important target for achieving a high level of independence in active daily life is the restoration of the hand function. Thus, the main objective in applying FES in such patients is to improve the hand function by creating a reliable and long lasting power grasp, or a smooth pulp-pinch grasp that is needed to manipulate small objects. Neuroprostheses for grasping are used to restore or improve grasping function in tetraplegic and stroke subjects. The patients that would benefit most from such a neuroprosthesis would be C5-7 quadriplegics. The available neuroprostheses for grasping enable the restoration of the two most frequently used grasping styles, the palmar and the lateral grasp (114) . The best known grasping neuroprostheses are the Freehand system (115) , Handmaster NMS (116) , Bionic Glove (117) , NEC FESMate system (118) , ETHZ-ParaCare neuroprosthesis (114) , the systems developed by Rebersek and Vodovnik (119) and the Belgrade Grasping system (120) . A recent overview can be found in (91) .

In 1997, FDA approved for clinical use the Freehand system (NeuroControl Corp., Cleveland) for restoration of the hand grasp in quadriplegics. The research and development process took 25 years, mostly carried out by the researchers at the Case Western Reserve University. The device consists of a stimulator and electrodes implanted in wrist and forearm muscles, a "joystick" controller implanted in the opposite shoulder, and an external processing unit (115) . The joystick-like sensor placed on the chest relaxes and tightens the hand as the shoulder moves back and forth. A detailed description of the prosthesis can be found in (121) . This prosthesis has been experimentally implanted in more than 130 people. With training, most patients with this device can open and close their hand in two different grasping movements and lock the grasp in place by moving their shoulder in different ways. A recent small clinical study with 6 patients (122) is a sign of increasing acceptance and application of the Freehand system outside USA. Researchers reported that all subjects were able to grasp, move and release more objects within the 30-s test period with the neuroprosthesis than without it. In 85% of the occasions, the six subjects expressed a preference for using the neuroprosthesis to perform these activities in daily living. Twelve months after rehabilitation, five of the six subjects still used the neuroprosthesis regularly. One of the main advantages of the Freehand system is that the time needed to put on (donning) and to take of (doffing) the system is significantly shorter compared to most surface stimulation FES systems. On the other hand, the Freehand system can be applied only 18-24 months after the injury and is only suitable for SCI subjects and not individuals suffering from stroke (114) .

The Bionic Glove is a hybrid system that utilizes a glove with FES electrodes. The detection of wrist extension stimulates finger flexion; therefore, the control of the device is directly related to the normal grasping sequence. Only patients who have sufficient wrist extension strength can use this type of device. The controls of the Freehand system and the Handmaster are not related to the normal movements included in the motor function, whereas the Bionic Glove uses mechanical sensors to detect wrist extension and the control is therefore directly coordinated by the grasping sequence. Bionic Glove can significantly improve independence in patients with C5-C7 spinal cord injury if their initial Functional Independence Measure and Quadriplegia Index of Function scores are 20% to 50% of the maximum values. The study is part of a multicenter clinical trial (123) .

The Handmaster is an orthosis for grasping with three pairs of surface stimulation electrodes. This system can be used to generate a grasping function in tetraplegic and stroke patients. Originally, this system was envisioned as an exercise and rehabilitation tool, but it is also used as a permanent prosthetic system.

One of the advantages of the Handmaster is that it is easy to put on and to take off. The Handmaster is predominately used as an exercise tool for stroke subjects and is commercially available in a limited number of countries.

An interesting attempt is the MeCFES system (myoelectrical controlled functional electrical stimulator) where the residual myoelectric signals from the paretic wrist extensor (m. extensor carpi radialis) are used to control stimulation of either the wrist extension (i.e., the same muscle) or thumb flexion. Initial results of six spinal cord lesion patients and one stroke patient show improvement of the wrist extension (124) .

With the exception of the Freehand and NEC-FES systems, all other neuroprostheses for grasping are FES systems with surface stimulation technology. Only the Freehand and Handmaster systems are currently available on the market while the other neuroprostheses are produced in laboratory environments.

All of the discussed neuroprostheses for grasping have demonstrated, in clinical trials in stroke and/or SCI subjects, improvement of the grasping function. These systems confirmed that the FES technology in principle could facilitate comfortable and secure grasp. However, the grasp strategies that can be provided with the existing neuroprostheses for grasping are very limited and can only be used for a restricted set of grasping and holding tasks.

In conclusion, it is too early to consider any of the existing systems, including Freehand and Handmaster, as success because of the small number of patients using the systems (about 150 for Freehand and a comparable amount for Handmaster).

Phrenic stimulation

Of those people who sustain a high cervical spinal cord injury, a substantial number initially require some form of mechanical ventilation of their lungs. Most of them will be able to breathe spontaneously with recovery, but 25% will remain dependent upon some form of ventilation support for the remainder of their lives. Most of these injuries occur in young, otherwise healthy people with a life expectancy of 20 years or more, assuming they have access to appropriate medical care. In patients with cervical lesions of the spinal cord, direct stimulation of the phrenic nerve can be applied for respiratory pacing. This is one of the earliest clinical applications of FES.

Diaphragm pacing is useful for conditions in which the brain stem respiratory centers provide little or no activation of the respiratory muscles, i.e. central hypoventilation syndrome, Arnold-Chiari malformation/brain stem dysfunction and high quadriplegia. Suitable patients are those with an intact phrenic nerve motor neuron pool in their cervical spinal cord on both sides. This can be checked by measuring the phrenic nerve conduction velocity and the diaphragm electromyogram (EMG). To do this phrenic nerves are stimulated by needle or surface electrodes in the neck and the diaphragm muscle EMG is recorded from electrodes placed low on the chest at the front. The patient wears an external radio-frequency (RF) transmitter over an implanted receiver, and a stimulating current is induced without the need for any percutaneous wires. The introduction of phrenic nerve pacing more than two decades ago by Dr William Glenn andassociates at Yale University has provided many ventilator-dependent tetraplegicpatients with freedom from mechanical ventilation (9, 125, 126) .

More than 1.200 phrenic nerve stimulator implantations have been performed throughout the world since 1968. Patients from several months of age to over 80 years of age have been successfully implanted and paced for long periods. Many patients have been successfully helped for more than 10 years; the longest period of pacing in a patient has been more than 20 years. In patients with ventilator-dependent quadriplegia, phrenic nerve pacing provides significant clinical advantages compared with mechanical ventilation. This technique, however, generally requires a thoracotomy with its associated risks; in-patient hospital stay and it carries some risk of phrenic nerve injury.

During the past decade, diaphragm pacing has also been attempted in small infants. The appropriate stimulation parameters are low stimulus frequency, short inspiration time and moderate respiratory rate. In a clinical trial in 33 pediatric patients, the mean time to failure was 56 months, which is acceptable for limited application of the pacemakers (127) . Among the benefits of the phrenic nerve pacing can be outlined:

Increased mobility of the patients

Improved sense of well-being and reduced anxiety due to elimination of fear of ventilator disconnection, elimination of ventilator tubing, elimination of ventilator noise, more physiological breathing and tracheostomy closure in some patients

Reduced incidence of respiratory tract infections

To recapitulate: one may consider the phrenic stimulator as a major improvement of the quality of life of the patient, which also reduces the overall costs for nursing of the patient.

Clinical trials / Evidence based medicine for FES

Efficacy of FES after stroke

Meta analysis showed that in post-stroke hemiparetic patients FES promotes recovery of muscle strength if included as part of the rehabilitation (based on the paretic muscle force measurements). The study was based on the results of clinical trials dating between 1978 and 1992 (128) . There is, however, more evidence for improvement of the motor control of the upper extremity after stroke (129) . The meta analysis was based on six randomized controlled trials on the therapeutic electrical stimulation on motor control and functional abilities. After stroke, up to 81% of the individuals develop shoulder subluxation, a condition frequently associated with poor upper limb function. FES on the shoulder muscles has been used in treatment of this condition. The results of seven (four early and three late) trials indicate when added to conventional therapy FES is beneficial early after stroke for prevention of shoulder subluxation (130). Burridge et al. in a randomized controlled trial in subacute single stroke patients measured the effect of the Odstock Dropped Foot Stimulator as a supplement to physiotherapy (100). The results of 32 subjects showed increase in walking speed and beneficial Physiological Cost Index, if the stimulator had been used.

Efficacy of FES in urological conditions

Stimulation of the sacral spinal nerves or the sacral spinal roots or the pelvic nerves can be used to restore bladder function in patients with voiding disorders. Several therapeutic techniques have become established in clinical urology as part of therapy at an increasing number of specialized centers.

As part of the treatment of different kinds of lower urinary tract dysfunctions, refractory to conservative treatment, Tanagho and Schmidt introduced sacral nerve neuromodulation in the 80s (131) . Neuromodulation is now carried out by stimulation of the S3 sacral nerve. In brief, an electrode is placed in S3 foramen. The electrode has to be brought into the ventral side of the opening in front of the S3 ventral ramus. This procedure is minimally invasive, as compared to sacral neurostimulation. The patient is sent home with an external pulse generator for a few days as part of an evaluation test. Responders are then implanted with a permanent sacral foramen implant and an implantable pulse generator (132) .

Since its early applications, neuromodulation has grown in popularity and the indications for this procedure are multiplying. Among them are now urge incontinence and sensory urgency, idiopathic chronic urinary retention, pelvic pain and interstitial cystitis (133). Complications in general are minimal and included electrode migration, electrode failure and pain at the implantable pulse generator site (134).

The sacral neurostimulation approach proposed in the 1980s by Brindley (135) consists of intradural implantation of a so-called book electrode to stimulate S2-S3 ventral roots, together with posterior S2-S4 rhizotomy. The primary purpose of the Brindley bladder stimulator is to improve bladder emptying, thereby eliminating urinary infection and preserving kidney function in suprasacral SCI patients. It also assists in defecation and enables male patients to have a sustained full erection. Nowadays the Brindley bladder stimulator is being developed by the company Finetech in UK.

Detrusor sphincter dysynergia

The suprasacral SCI patients develop detrusor sphincter dyssinergia, which imposes serious health risk. The results from the first 500 patients implanted with the Brindley stimulator show usage in 424 after a mean follow up of 4 years. (136) . Another study reported 38 patients with a complete spinal cord lesion (137) . During the follow-up period, ranging up to 12 years all patients had increased bladder capacity and reduced residual urine volumes; 31 patients were continent; 29 males could achieve a sustained full erection; 27 patients used the implant for bowel function. The long long-term favorable effects of the stimulator were also confirmed in follow-up studies on urodynamics (138) and cost-effectiveness (139) in 52 patients.

Overactive bladder

The overactive bladder comprises a spectrum of conditions ranging from urgency-frequency syndrome to urge incontinence. The incidence of the overactive bladder increases with age, with a prevalence exceeding 4% to 5.5% of the population. The efficacy of the neuromodulation therapy was evaluated in a prospective 12-center study in Europe, Canada, and the United States conducted form 1993 to 1999 with the InterStim system (Medtronic, Inc., Minneapolis, Minnesota). The study enrolled 184 patients for urinary urge incontinence, 220 for urgency-frequency, and 177 for retention for a total of 581 patients (140) .

Urge Incontinence and Sensory Instability

Neuromodulation is used in patients with urge incontinence and with therapy-resistant idiopathic detrusor instability, where in a follow-up trial results showed clinically significant improvement of the quality of life (133) . In patients with detrusor hyperactivity implantable neuroprosthetic devices also lead to improvement in the urodynamic parameters (141) . Out of 20 patients with urge urinary incontinence, presented by Thon and colleagues (2003) (142) , 17 showed an improvement of more than 50% compared to the baseline, which persisted for more than a year of follow-up. Elabbady and colleagues (1994) (143) presented their results in 9 patients with urgency frequency and/or urge incontinence: frequency improved by 73%, urgency by 42% and incontinence by 50%. However, the number of patients is small to derive statistically valid conclusions for the success rate of the operation.

On the other hand, the outcome of the multi-center trial is promising. Results demonstrate that after 3 years, 59% of 41 urinary urge incontinent patients showed greater than 50% reduction in leaking episodes per day with 46% of patients being completely dry. After 2 years, 56% of the urgency-frequency patients showed greater than 50% reduction in voids per day (140) .

Idiopathic Non-Obstructive Chronic Urinary Retention

The results for the idiopathic urinary retention of the multi-center trial were also separately reported in (144) . A total of 177 patients with urinary retention refractory to standard therapy were enrolled in the study. Compared to the control group, patients implanted with the InterStim system had statistically and clinically significant reductions in the catheter volume per catheterization. Successful results were achieved in 83% of the implant group with retention compared to 9% of the control group at 6 months. Temporary inactivation of sacral nerve stimulation therapy resulted in a significant increase in residual volumes but effectiveness of sacral nerve stimulation was sustained through 18 months after implant. In another study, Thon and colleagues (2003) (142) reported that from 33 patients with chronic urinary retention implanted permanently with neuroprosthesis 23 showed a long-lasting significant improvement, but in the remaining 10 the improvement did not reach 50% compared to baseline. In 7 patients with chronic retention, Vapnek and Schmidt (2003) reported for success in 5 cases (145) and Elabbady and colleagues (1994) (143) presented success in 8 of 8 cases. Results of all the presented studies demonstrate that sacral nerve stimulation is effective for restoring voiding in patients with retention who are refractory to other forms of treatment.

Pelvic pain and discomfort

Pelvic pain or discomfort is a very common symptom associated with other storage or voiding dysfunctions. In the available literature on sacral root neuromodulation, associated pelvic pain has improved from 85% to 90% when postimplant status was compared to baseline (134, 142).

In conclusion: The correlation between the clinical outcome and the urodynamic test results is poor. The results of the presented studies on neuromodulation show that 40% of the selected patients do not qualify for the procedure and the effect of treatment is not enduring: voiding dysfunctions return as soon as the neuroprosthesis is switched off. It is thought that the effect is based on antidromic stimulation of the inhibitory neurons in the spinal cord. Due to the fact that the stimulators are voltage-controlled so far, the amount of injected current into the tissues is unknown and the fibers which are stimulated are also not known (may be both myelinated or non-myelinated) and also the direction of stimulation (ortho- antidromic). More basic and clinical research needs to be performed before this treatment can be introduced as a routine procedure in patients with serious voiding dysfunction refractory to conservative measures (146) .

Experimental and theoretical research

Experimental research

Visual prosthesis: blind eye vs. normal brain

Blindness can result when any step of the optical pathway—the optics, the retina, the optic nerve, visual cortex, or other cortical areas involved in the processing of vision — sustains damage. In Germany, 17.000 patients become blind every year for whom there is no effective treatment or cure (1).

Public discussion of electricity's effect on vision dates to 1751, when it was addressed by Benjamin Franklin following his celebrated kite-and-key experiment. Despite some advocates, the idea of treating blindness through electrical stimulation did not catch on. In the last 30 years extensive experimental research for development of visual prostheses has been performed. In 1967-68 the experiments of Brindley and Lewin (14) showed the feasibility of the long-term interface with the visual system. The approach of the cortical prostheses has its foundations in the observations of the visuotopic organization of electrically evoked phosphenes in the occipital cortex. This have led a number of investigators to propose that electrical stimulation of visual cortex via arrays of electrodes might provide the profoundly blind with a limited form of functional vision. Several groups investigated the stimulation with cortical surface electrodes – the so called “cortical prostheses” of Brindley, Dobelle (1974) (147) ; cortical penetrating electrodes of Bartlett and Doty (1980) (148) and of Schmidt and Hambrecht (1996) (149) . However influential it is, so far this idea has very little practical progress. The neurons in the cortex represent textures, depth (displacements), angles and brightness /colors of the object. The perception of images is not based on pixelation; therefore, all the bitmap-based stimulation approaches are doomed to fail. On the other hand, the cortical stimulation approach may provide the only therapeutic approach for individuals with non-functional retinas and/or optic nerves.

Recently Dobelle (2000) (150) reported for the development of a visual prosthesis providing a sort of "artificial vision" to a blind volunteer by connecting a digital video camera, computer and associated electronics to the visual cortex of his brain. As an alternative to the cortical stimulation, two other approaches have been also investigated.

The idea to realize a visual prosthesis by stimulating the optic nerve was conceived by Mortimer and Veraart (1998) (151) at the beginning of the 1990's. The first implantation was in 1998 and the patient was able to localize single bright spots of light, but high spatial resolution cannot be expected with such a stimulation arrangement. This prosthesis may show promise if an ultra-high electrode count array of intraneural electrodes are implanted in the optic nerve for providing better selectivity of stimulation. Although ingenious, the approach does not take into account so far the information processing steps taking places in the retinal ganglionic cells.

At the end of the 1980's, an entirely new approach was undertaken by the teams of M. Humayun, of John Hopkins University (149) and that of J. Rizzo of Harvard University (156) , in association with the Massachusetts Institute of Technology. They developed an implantable electrode array for the retina itself, further referred to as retinal implant, in order to stimulate the retinal ganglionic cells (152) , whose extensions form the optic nerve. This prosthesis is referred to as "epiretinal implant". Another kind of retinal implant is the subretinal implant developed by Chow and colleagues (1997) (153) in Chicago and Zrenner in Tubingen (1997) (154) .Both teams (155, 156), have stimulated the retina of blind patients with epiretinal electrodes that were transiently inserted into the eye through a scleral opening. Both groups reported a sensation of light patterns by the patients, but perception of geometric patterns was reported in only a few instances. Recently a permanent implantation has been made in a blind volunteer (157) .

All results in the field in the past 30 years are still far from true object recognition. However, they do demonstrate the feasibility of generating perception of light patterns in blind people.

Spinal Cord Stimulation in motor disorders

Apart from the pain suppression of chronic intractable pain, peripheral vascular disease and angina pectoris, spinal cord stimulation has been employed in motor disorders for control of spasticity in SCI patients. Attempts in this direction have been carried out in the 1980s for the first time by Richardson et al. (1979) (161) . Recently Pinter et al. (2000) (162) stimulated the L2-L3 dorsal roots of the spinal cord in SCI patients. Results demonstrated reduction of the muscle hypertonia of the lower extremities.

A possible future application of SCS is to contribute to the restoration of the motor functions of the lower extremity (158, 159) . The reasoning behind this idea is that if we are able to control the Central Pattern Generator (CPG) in the motor spinal cord we circumvent many of the selectivity problems pointed out so far. There is accumulating evidence for the existence of CPG of the locomotion in primates and men (160). This requires a Multi Electrode Array to be placed on the spinal cord such that it is able to stimulate the musculotopically-organized motor pools of the lower limbs.

Theoretical Modeling studies

As part of the design of different types of electrodes for stimulation a substantial amount of theoretical and experimental research on selectivity has been performed. The electrical behavior of the myelinated nerve fiber can be represented by a simple cable network. The first to introduce this model was McNeal (1976) (27) in order to calculate how a stimulation-induced extracellular field affects nodal transmembrane voltages. According to this model, the node of Ranvier, which is closest to the cathode, will be excited first when the stimulus current is sufficiently high. The results of this theoretical approach are in accordance with the general observation that the threshold stimulus of a nerve fiber excitation is smallest near the cathode and rises with the increase of the distance (23) .

Research on spatially-selective nerve stimulation

The feasibility of selective activation of peripheral nerve fascicles was demonstrated by McNeal and Bowman (1985) (31) , who found that with proper fit and positioning a single circumneural sleeve with multiple electrode contacts could control selectively the activation of two antagonist muscle groups innervated by a common nerve trunk. In their study, Sweeny et al. (1990) (42) performed numerical modeling and experimental testing of a nerve cuff technique for selective stimulation of superficial peripheral nerve trunk regions. Two basic electrode configurations ("snug" cuff monopolar and tripolar longitudinally aligned dots) have been considered. Both modeling and experimentation suggested that longitudinally aligned tripolar dot electrodes on the surface of a nerve trunk would restrict excitation to superficial nerve trunk regions more successfully than monopolar dot electrodes would do. Transverse anodal "steering" improved the spatial selectivity of both monopolar and tripolar electrode configurations (42) .

With a sufficient number of electrodes within a so-called “cuff electrode”, high selectivity of stimulation can be achieved by either longitudinal or transversal currents produced by appropriately switched electrodes (41) .

Goodall et al. (1996) (45) found that transverse current from an anode positioned opposite the stimulating cathode improved spatial selectivity, and position selectivity was enhanced when the ratio of transverse current to longitudinal current was increased. Based on computer modeling Deurloo et al. (2003) concluded that transverse steering provides more selective stimulation than longitudinal steering (44) .

Figure 7. Transverse section of the 3D volume conductor model of a monofascicular nerve surrounded by an insulating cuff with an electrode contact on its inner surface.

B Recruitment contours showing excitation regions in the nerve trunk. c- cathode, a –anode; a: central cathode b: longitudinal tripole + additional anode c: transverse tripole.

Courtesy: K Deurloo, from (96) A p. 15 and B p.30

Selectivity was dependent on the relative location of the electrode contacts and the nerve fascicles, as well as the size and relative spacing of neighboring fascicles (43) .

Deurloo et al. (1998) (163) performed theoretical research on how to improve the selectivity performance of multi-contact cuff electrodes for stimulation of peripheral nerves. As combination of controlled fiber size and spatial selectivity seems hard to achieve, the focus was primarily on spatial selectivity. They found that the transverse tripole is the only configuration that maximizes activation selectivity for a small (cylindrical) bundle of fibers in the periphery of a monofascicular nerve trunk (163) (see also Fig 7). Inverse recruitment was less pronounced than for the other configurations. Therefore, they recommended transverse tripolar stimulation since it did not change the shape of the recruitment contours, despite the lowering of the excitation threshold, which might occur in chronic implantation fibrosis on the implantation site.

However, in acute animal experiments, where the recruitment characteristics of muscle selective nerve stimulation by a multi-contact nerve cuff electrode was studied, the results showed that only in a few cases transverse bi- and tripolar stimulation provided a better selectivity than monopolar stimulation (164) . In accordance with the results of the modeling studies, bi- and tripolar stimulation required higher stimulus currents than monopolar stimulation, whereas maximum recruitment and slopes of recruitment curves were lower. Due to the variability in the number and size of the fascicles and their position in this nerve, sufficient reproducibility for the selectivity could not be obtained.

The theoretical research performed by Rutten et al. (1991) (56) and Yoshida, and Horch (1993) (49) suggested that intraneural electrodes should give better spatial selectivity than cuff electrodes. In this line of reasoning Rutten et al. (1999) (165) argue that the best way to control individual fibers is to stimulate in close proximity of the nodes of Ranvier, which implies use of intrafascicular electrodes. However, the distribution of the nodes of Ranvier in the peripheral nerves makes the stimulation a probabilistic process. The experimental studies show that, despite the theoretical expectations, when the topography of the motor nerve fibers is unknown, the implanted intraneural electrode may simply miss the target fiber population. Several attempts for such stimulation have been made so far with 1D arrays. The obtained information gave insight on how to improve the design (shape, distance between the electrodes) of the later on produced 3D arrays.

In acute nerve implantation experiments eight 5- to 24-wire-MEAs were used in order to investigate whether the electrodes could selectively stimulate single motor units. The results revealed partial blocking of neural conduction, similar to that reported with microneurographic insertion with single needles (166) . The eight arrays were capable of evoking threshold forces selectively with an average efficiency of 0.81 (or 81%)

Frieswijk et al. (1998) (167) performed animal experiments and model simulations of monopolar, intrafascicular nerve stimulation in order to study force-current relationships (recruitment curves). They found that the conductivity of the extraneural medium is of prime importance to the resulting recruitment curves: an insulating extraneural medium generally leads to steeper curves with lower threshold currents than a well-conducting extraneural medium. The statistical comparison of experimental and model results suggested clustering of the motor fibers, originating from a single muscle, within the same fascicle. This clustering manifested itself mainly by an increased spread in threshold currents, as opposed to the situation where the fibers are distributed uniformly throughout the entire fascicle.

In conclusion, the insufficient neuroanatomical knowledge of the spatial organization and topography of fibers in the peripheral nerve hampers the definitive choice for the design of the electrode.

Research on fiber size selective stimulation

In myelinated axons, there is a positive correlation between the internodal distance and the axonal diameter. Thus, the large fibers (A, B) have nodes further apart than the small fibers. As a result, in an electric field the larger fibers have a larger potential difference between adjacent nodes. This leads to a lower threshold of the large fibers in comparison with the small ones. Therefore, during stimulation, the large myelinated fibers fire first. Unfortunately, this is the reverse of the natural order of recruitment, and it results in fast development of muscle fatigue. By activating/blocking stimuli delivered by multiple cuff electrodes the order of recruitment can be reversed, but the fibers still can not be chosen (168) . Better fiber size selectivity can be achieved by intraneural stimulation, as outlined by Rutten et al. (1999) (165) .

The order in which the nerve fibers in a peripheral nerve or a spinal root are excited by a stimulus pulse is predominantly related to both the fiber diameter and the distance between the fiber and the cathode (169) . In the conducting media around the electrode, the current density is inversely proportional to 2^nd- 3^rd power of the distance. Accordingly, the threshold current is increased at the same rate as shown empirically (23) . In clinical applications, due to the generally limited amplitude range, only the large fibers will be recruited. For example in spinal cord stimulation, the maximal therapeutic amplitude should not exceed 170% of the paresthesia threshold perception. Calculations show that the nerve fibers smaller than 9um in diameter will not be recruited. Another relevant aspect of the recruitable large fibers is their density within the fascicle (169) .

Fang and Mortimer (1991) (170, 171) studied selective activation of small fibers without activating larger fibers in the same nerve trunk. In the proposed nerve stimulation system, quasitrapezoidal-shaped current pulses were delivered through a tripolar cuff electrode to effect differential block by membrane hyperpolarization. The quasitrapezoidal-shaped pulses with a square leading edge, a 350 microsecond(s) plateau, and an exponential trailing phase ensured the block of propagating action potentials and prevented the occurrence of anodal break excitation. The tripolar cuff electrode restricted current flow inside the cuff and thus eliminated the undesired nerve stimulation due to a "virtual cathode." The subsequent animal experiments confirmed that larger alpha motor axons could be blocked at lower current levels than smaller alpha motor axons, and that all alpha fibers could be blocked at lower current levels than gamma fibers, and the blocking threshold correlated with the fiber diameter Fang and Mortimer (1991 b) (171). For the same purpose, van Bolhuis et al. (2001) (172) suggested the use of two pulse generators independently supplying short supramaximal cathodal stimulating pulses (0.5 ms) and long subthreshold cathodal inactivating pulses (1.5 s) to the sciatic nerve. Results showed that propagation of action potentials was selectively blocked in nerve fibers of different diameters by adjusting the strength of the inactivating current (172) .

It has been shown that cathodal pre-pulses inverse the recruitment order

(173, 174) . Grill and Mortimer (1997) (173) demonstrated that subthreshold membrane depolarization generated a transient decrease in neural excitability and thus an increase in the threshold for stimulation by a subsequent stimulus pulse. When a depolarizing stimulus pulse was applied immediately after the subthreshold depolarization, nerve fibers far from the electrode could be stimulated without stimulating fibers close to the electrode. Thus subthreshold depolarizing pre-pulses allowed selective stimulation of nerve fibers far from the electrode

In a realistic model of a nerve fiber surrounded by a cuff electrode Deurloo et al. (2001) (174) showed that it is also possible to stimulate small fibers without exciting large ones. The applied model requires that, in the case of monopolar stimulation with a cuff electrode, the cuff length should not exceed twice the internodal length of the fibers to be blocked. Similarly, the distance between cathode and anodes should not exceed the internodal length of these fibers when tripolar stimulation is used (174) .

Conclusions

The principal requirements to any substituting structure are features mimicking some of the biological functions of nerves and replacing these functions depending on the scope of implantation. Profound knowledge about the development, functional and structural organization of the nervous system is a prerequisite for any attempt to establish meaningful recording or stimulation in order to substitute for a given function or modality.

A number of proposed ideas for electronic implants are based, however, on a more "engineer-like" way of thinking rather than considering the complexity of the biological systems. The necessary for the engineering development reductionist approach often substitutes basic problems for secondary ones. For example, the concept of spatial selectivity in stimulation is often substituted for fascicle selectivity, which equalizes the fiber representation of a particular muscle contraction (dynamic view) to a coincidentally delimited anatomical region along the course of a nerve (static view).

The presented overview of the FES applications shows that despite the substantial amount of research performed in the field which led to successful development of cardiac and phrenic pacemakers and cochlear prostheses, a number of fundamental scientific problem still remains to be solved before we see a comparable degree of effectiveness and penetration in common medical practice for the locomotion neuroprostheses and urological FES appliances.

Although FES of locomotion gains clinical acceptance, it is still in its experimental phase of development. There are several interrelated, but different, issues

- Effectiveness and reliability of the devices

- Technical support

- The place of the devices in the complex treatments of the CVA, SCI, multiple sclerosis etc.

Despite the significant technical progress achieved in the last 10 to 15 years in the FES field, there is a consensus that these systems are not sufficiently advanced and that they need further development. Complexity of human motion greatly diminishes the area of the animal experimentation in this field and puts the stress on computer modeling. This complexity, however, makes the leap from a model to a patient even more difficult than from an animal model to human, which one could see in the performance on many implantable FES systems. This has only been done in limited number of cases, which in turn diminishes the validity of the obtained results and leads to case descriptions rather than generalizations. This in turn increases the interval between trial and error (the most prominent approach in experimentation) and as a result slows down the research and development process for new applications.

Nevertheless, the present FES treatments combined with conventional occupational and physical therapy still remain the most promising approach in rehabilitating SCI patients and stroke patients. The need of training, which is seen by some as a shortcoming of FES (97) , in our opinion may be turned into advantage if it is integrated in the overall treatment and rehabilitation process.

The mechanism of action of electrical neuromodulation is still poorly understood, which leaves it in a rather empirical state of operation. The Brindley neurostimulation method is a real FES application. Until now hundreds of patients have been treated successfully with the Brindley approach in bladder voiding disorders. However, the posterior rhizotomy, which is required, is a rather crude approach towards blocking the spinal reflexes. Nevertheless, FES is effective in improving the bladder function in the overactive bladder states and in SCI patients it gives long-term favorable results, which makes it a viable therapeutical option for the indicated groups of patients.

The research performed in the field of the visual prostheses showed the complexity of the information processing by the (human) visual system. Even the modest goals of the projects so far are ambitious compared with the information content of the reported visual phenomena. We would like to finish with a quotation from Dobelle (150) , which applies in the end to any prosthetic treatment: “Development of implanted medical devices such as this artificial vision system progresses in three stages. First there is speculation, then there is hope, and finally there is promise.”

References

1 . Moe JH, Post HW. Functional electrical stimulation for ambulation in hemiplegia. J Lancet 1962; 82: 285-288.

2. Dutrochet RJH. Recherches anatomiques et physiologiques sur la structure intime des animaux et vegétax et sur leur mobilité. Paris: Bailliére, 1824.

3. Remak R. Observationes anatomicae et microscopiscae de systematis nervosi structura. Berolini: Reimerans, 2003.

4. Rattay F. Electrical nerve stimulation: theory, experiments and applications. Wien: Springer-Verlag , 1990.

5. Gold GR. Cardiac pacing-from then to now. Medical Intsrumentation 1984; 18: 15-21.

6. Greatbatch W. Origins of the implantable cardiac pacemaker. J Cardiovasc.Nurs. 1991; 5: 80-85.

7. Chardack WM, Gage AA, Greatbatch W. Experimental observations and clinical experiences with the correction of complete heart block by an implantable, self-contained pacemaker. Trans.Am.Soc.Artif.Intern.Organs 1961; 7: 286-295.

8. Sarnoff SJ, Gaensler EA, Maloney JV. Electrophrenic respiration: the effectiveness of contralateral ventialtion during activity of one phrenic nerve. J.Thoracic.Surg. 1950; 19: 929.

9. Glenn WW, Hageman JH, Mauro A, Eisenberg L, Flanagan S, Harvard BM. Electrical stimulation of excitable tissue by radiofrequency transmission. Ann Surg 1964; 160: 338-350.

10. Clark GM, Tong YC. Cochlear Prostheses. Melbourne: Churchill Livingstone, 1990.

11. Djourno A, Eyries C. Prothèse auditive par excitation électrique a distance du nerf sensoriel a l'aide d'un bodinage inclus a demeure. Presse Med 1957; 35: 14-17.

12. Foerster O. Beitrage zur Pathophysiologie der Sehbahn und der Sehsphare. J.Psychol.Neurol., Lpz 1929; 39 : 463-485.

13. Krieg W. Functional Neroanatomy. New York: Blakiston, 1953: 207-8.

14. Brindley GS, Lewin WS. The visual sensations produced by electrical stimulation of the medial occipital cortex. J Physiol 1968; 194: 54-5P.

15. Liberson WT, Holmquest HJ, Scot D, DOW M. Functional electrotherapy: stimulation of the peroneal nerve synchronized with the swing phase of the gait of hemiplegic patients. Arch.Phys.Med.Rehabil. 1961; 42: 101-105.

Hambrecht FT. The History of Neural Stimulation and Its Relevance to Future Neural Prostheses. In: Agnew WF, McCreery DB, editors. Neural Prostheses: Fundamental Studies. Englewood Cliffs: Prentice Hall, 1990: 1-25.

17. Robblee LS, Rose TL. Electrochemical guidelines for selection of protocols and electrode materials for neural stimulation. In: Agnew WF, McCreery DB, editors. Neural Prostheses: Fundamental studies. Englewood Cliffs: Prentice Hall, 1990: 26-66.

18. Bhadra N, Peckham PH. Peripheral nerve stimulation for restoration of motor function. J Clin.Neurophysiol. 1997; 14: 378-393.

. Morton SL, Daroux ML, Mortimer JT. The role of oxygen reduction in electrical stimulation of neural tissue . J.Electrochem.Soc. 1994; 141 : 122-130.

20. Lilly JC, Hughes JR. Brief, noninjurious electric waveform for stimulation of the brain. Scince 1955; 121: 468-469.

21. McCreery DB, Agnew WF, Yuen TG, Bullara LA. Damage in peripheral nerve from continuous electrical stimulation: comparison of two stimulus waveforms. Med Biol.Eng Comput. 1992; 30: 109-114.

22. McCreery DB, Agnew WF, Yuen TG, Bullara LA. Relationship between stimulus amplitude, stimulus frequency and neural damage during electrical stimulation of sciatic nerve of cat. Med Biol.Eng Comput. 1995; 33: 426-429.

23. Ranck JB, Jr. Which elements are excited in electrical stimulation of mammalian central nervous system: a review. Brain Res. 1975; 98: 417-440.

24. Wee AS, Leis AA, Kuhn AR, Gilbert RW. Anodal block: can this occur during routine nerve conduction studies? Electromyogr.Clin.Neurophysiol. 2000; 40: 387-391.

25. Wee AS. Anodal excitation of intact peripheral nerves in humans. Electromyogr.Clin.Neurophysiol. 2001; 41: 71-77.

26. Hodgkin AL, Huxley AF. A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol 1952; 117: 500-544.

27. McNeal DR. Analysis of a model for excitation of myelinated nerve. IEEE Trans.Biomed.Eng 1976; 23: 329-337.

28. Talonen PP, Baer GA, Hakkinen V, Ojala JK. Neurophysiological and technical considerations for the design of an implantable phrenic nerve stimulator. Med Biol.Eng Comput. 1990; 28: 31-37.

29. Heiduschka P, Thanos S. Implantable bioelectric interfaces for lost nerve functions. Prog.Neurobiol. 1998; 55: 433-461.

30. Rutten WL. Selective electrical interfaces with the nervous system. Annu.Rev.Biomed.Eng 2002; 4: 407-452.

31. McNeal DR, Bowman BR. Selective activation of muscles using peripheral nerve electrodes. Med Biol.Eng Comput. 1985; 23: 249-253.

32. Marsolais EB, Kobetic R. Functional walking in paralyzed patients by means of electrical stimulation. Clin.Orthop. 1983; 30-36.

33. Wuolle KS, Van Doren CL, Thrope GB, Keith MW, Peckham PH. Development of a quantitative hand grasp and release test for patients with tetraplegia using a hand neuroprosthesis. J Hand Surg.[Am.] 1994; 19: 209-218.

34. Marsolais EB, Scheiner A, Miller PC, Kobetic R, Daly JJ. Augmentation of transfers for a quadriplegic patient using an implanted FNS system. Case report. Paraplegia 1994; 32: 573-579.

35. Loeb GE, Peck RA, Moore WH, Hood K. BION system for distributed neural prosthetic interfaces. Med Eng Phys. 2001; 23: 9-18.

36. Cameron T, Loeb GE, Peck RA, Schulman JH, Strojnik P, Troyk PR. Micromodular implants to provide electrical stimulation of paralyzed muscles and limbs. IEEE Trans.Biomed.Eng 1997; 44: 781-790.

37. Naples GG, Mortimer JT, Yuen TGH. Overview of peripheral nerve electrode design and implantation. In: Agnew WF, McCreery DB, editors. Neural Prostheses: Fundamental studies. Englewood Cliffs: Prentice Hall, 1990: 107-45.

38. Naples GG, Mortimer JT, Scheiner A, Sweeney JD. A spiral nerve cuff electrode for peripheral nerve stimulation. IEEE Trans.Biomed.Eng 1988; 35: 905-916.

39. Kim JH, Manuelidis EE, Glenn WW, Fukuda Y, Cole DS, Hogan JF . Light and electron microscopic studies of phrenic nerves after long-term electrical stimulation. J Neurosurg. 1983; 58: 84-91.

40. Klepinski, J. R. Implantable neural electrode. 1994. USA.

41. Veraart C, Grill WM, Mortimer JT. Selective control of muscle activation with a multipolar nerve cuff electrode. IEEE Trans.Biomed.Eng 1993; 40: 640-653.

42. Sweeney JD, Ksienski DA, Mortimer JT. A nerve cuff technique for selective excitation of peripheral nerve trunk regions. IEEE Trans.Biomed.Eng 1990; 37: 706-715.

43. Grill WM, Jr., Mortimer JT. Quantification of recruitment properties of multiple contact cuff electrodes. IEEE Trans.Rehabil.Eng 1996; 4: 49-62.

44. Deurloo K, Holsheimer J. Fascicular Selectivity in Transverse Stimulation with a Nerve Cuff Electrode: A Theoretical Approach. Neuromodulation 2003; 6: 258-269.

45. Goodall EV, de Breij JF, Holsheimer J. Position-selective activation of peripheral nerve fibers with a cuff electrode. IEEE Trans.Biomed.Eng 1996; 43: 851-856.

46. Loeb GE, Peck RA. Cuff electrodes for chronic stimulation and recording of peripheral nerve activity. J Neurosci.Methods 1996; 64: 95-103.

47. Tarver WB, George RE, Maschino SE, Holder LK, Wernicke JF. Clinical experience with a helical bipolar stimulating lead. Pacing And Clinical Electrophysiology: PACE 1992; 15: 1545-1556.

48. Nannini N, Horch K. Muscle recruitment with intrafascicular electrodes. IEEE Trans.Biomed.Eng 1991; 38: 769-776.

49. Yoshida K, Horch K. Selective stimulation of peripheral nerve fibers using dual intrafascicular electrodes. IEEE Trans.Biomed.Eng 1993; 40: 492-494.

50. BeMent SL, Wise KD, Anderson DJ, Najafi K, Drake KL. Solid-state electrodes for multichannel multiplexed intracortical neuronal recording. IEEE Trans.Biomed.Eng 1986; 33: 230-241.

51. Rutten WL, Frieswijk TA, Smit JP, Rozijn TH, Meier JH. 3D neuro-electronic interface devices for neuromuscular control: design studies and realisation steps. Biosens.Bioelectron. 1995; 10: 141-153.

52. Nordhausen CT, Maynard EM, Normann RA. Single unit recording capabilities of a 100 microelectrode array. Brain Res. 1996; 726: 129-140.

53. Branner A, Normann RA. A multielectrode array for intrafascicular recording and stimulation in sciatic nerve of cats. Brain Res.Bull. 2000; 51: 293-306.

54. Meyer, J-U., Blau, C., and Stieglitz, Th. Multikontaktierung von Nerven mit Mikrostrukturen. Spektrum D.Wissensch. 98-102. 1995.

55. Smit, J. P. A. Selective Motor Stimulation using Endoneural Prostheses. 1-24. 1996. Twente University.

56. Rutten WL, van Wier HJ, Put JH. Sensitivity and selectivity of intraneural stimulation using a silicon electrode array. IEEE Trans.Biomed.Eng 1991; 38: 192-198.

57. Fried K, Frisen J, Mozart M. De- and regeneration of axons after minor lesions in the rat sciatic nerve. Effects of microneurography electrode penetrations. Pain 1989; 36: 93-102.

58. Rice ASC, Andreev NY, McMahon SB. The consequences of microneurography electrode-induced injury of peripheral nerves observed in the rat and man. Pain 1994; 59: 385-393.

59. Rice AS, McMahon SB. Peripheral nerve injury caused by injection needles used in regional anaesthesia: influence of bevel configuration, studied in a rat model. British Journal Of Anaesthesia 1992; 69: 433-438.

60. Rice ASC, McMahon SB, Wall PD. The electrophysiological consequences of electrode impalement of peripheral nerves in the rat. Brain Research 1993; 631: 221-226.

61. Bowman BR, Erickson RC. Acute and chronic implantation of coiled wire intraneural electrodes during cyclical electrical stimulation. Annals Of Biomedical Engineering 1985; 13: 75-93.

62. Hallin RG, Wiesenfeld-Hallin Z, Duranti R. Percutaneous microneurography in man does not cause pressure block of almost all axons in the impaled nerve fascicle. Neuroscience Letters 1986; 68: 356-361.

63. Lefurge T, Goodall E, Horch K, Stensaas L, Schoenberg A. Chronically implanted intrafascicular recording electrodes. Annals Of Biomedical Engineering 1991; 19: 197-207.

64. Stanic U, Bajd T, Valencic V, Kljajic M, Acimovic R. Standardization of kinematic gait measurements and automatic pathological gait pattern diagnostics. Scand.J Rehabil.Med 1977; 9: 95-105.

65. Petrofsky JS, Phillips CA. Discharge characteristics of motor units and the surface EMG during fatiguing isometric contractions at submaximal tensions. Aviat.Space Environ.Med 1985; 56: 581-586.

66. Williams DF. Definitions in Biomaterials. Progress in Biomedical Engineering. Amsterdam: Elsevier, 1987: 54.

67. Bollen, L. S. and Svendsen, O. Regulatory Guidelines For Biocompatibility Safety Testing. Medical Plastics and Biomaterials , 16. 1-5-1997.

68. Salthouse TN. Cellular enzyme activity at the polymer-tissue interface: a review. J Biomed.Mater.Res. 1976; 10: 197-229.

69. Ratner BD, Hoffman A, Lemons JE, Schoen FJ. Biomaterials Science - An Introduction to Materials in Medicine. New York : Academic Press, 1996.

70. Agnew WF, McCreery DB. Neural Prostheses: Fundamental studies. Englewood Cliffs: Prentice Hall, 1990.

71. Vaudaux PE, Francois P, Proctor RA, McDevitt D, Foster TJ, Albrecht RM et al. Use of adhesion-defective mutants of Staphylococcus aureus to define the role of specific plasma proteins in promoting bacterial adhesion to canine arteriovenous shunts. Infect.Immun. 1995; 63: 585-590.

72. Wassall MA, Santin M, Peluso G, Denyer SP. Possible role of alpha-1-microglobulin in mediating bacterial attachment to model surfaces. J Biomed.Mater.Res. 1998; 40: 365-370.

73. AAMI. Association for the Advancement of Medical Instrumentation American National Standard for Implantable Peripheral Nerve Stimulators. 1984. ANSI/AAMI NS15-1984.

74. Romero, E. Feasibility of Long-term Nerve Stimulation with the Self-sizing Spiral Cuff Electrode: Histological and Physiological Study. 1-207. 2000. Université catholoque de Louvain.

75. Grill WM, Mortimer JT. Electrical properties of implant encapsulation tissue. Ann Biomed.Eng 1994; 22: 23-33.

76. Naples GG, Mortimer JT, Scheiner A, Sweeney JD. A spiral nerve cuff electrode for peripheral nerve stimulation. IEEE Trans.Biomed.Eng 1988; 35: 905-916.

77. Agnew WF, McCreery DB, Yuen TG, Bullara LA. Histologic and physiologic evaluation of electrically stimulated peripheral nerve: considerations for the selection of parameters. Ann Biomed.Eng 1989; 17: 39-60.

78. Grill WM, Mortimer JT. Neural and connective tissue response to long-term implantation of multiple contact nerve cuff electrodes. J Biomed.Mater.Res. 2000; 50: 215-226.

79. McNeal DR, Waters R, Reswick J. Experience with implanted electrodes. Neurosurgery 1977; 1: 228-229.

80. Koller R, Girsch W, Liegl C, Gruber H, Holle J, Losert U et al. Long-term results of nervous tissue alterations caused by epineurial electrode application: an experimental study in rat sciatic nerve. Pacing Clin.Electrophysiol. 1992; 15: 108-115.

81. Girsch W, Koller R, Gruber H, Holle J, Liegl C, Losert U et al. Histological assessment of nerve lesions caused by epineurial electrode application in rat sciatic nerve. J Neurosurg. 1991; 74: 636-642.

82. Agnew WF, McCreery DB, Bullara L, Yuen TG. Effects of Prolonged Electrical Stimulation of Peripheral nerve. In: Agnew WF, McCreery DB, editors. Neural Prostheses: Fundamental Studies. Englewood Cliffs: Prentice Hall, 1990: 147-68.

83. Liang F, Jones EG. Peripheral nerve stimulation increases Fos immunoreactivity without affecting type II Ca2+/calmodulin-dependent protein kinase, glutamic acid decarboxylase, or GABAA receptor gene expression in cat spinal cord. Exp.Brain Res. 1996; 111: 326-336.

84. Nakamura M, Rosahl SK, Alkahlout E, Gharabaghi A, Walter GF, Samii M. C-Fos immunoreactivity mapping of the auditory system after electrical stimulation of the cochlear nerve in rats. Hearing Research 2003; 184: 75-81.

85. Lanahan A, Worley P. Immediate-Early Genes and Synaptic Function. Neurobiology of Learning and Memory 1998; 70: 37-43.

86. Illing RB, Michler SA, Kraus KS, Laszig R. Transcription Factor Modulation and Expression in the Rat Auditory Brainstem Following Electrical Intracochlear Stimulation. Experimental Neurology 2002; 175: 226-244.

87. Molander C, Hongpaisan J, Grant G. Changing pattern of c-FOS expression in spinal cord neurons after electrical stimulation of the chronically injured sciatic nerve in the rat. Neuroscience 1992; 50: 223-236.

88. Illing RB. Activity-dependent plasticity in the adult auditory brainstem. Audiol.Neurootol. 2001; 6: 319-345.

89. Platenik J, Kuramoto N, Yoneda Y. Molecular mechanisms associated with long-term consolidation of the NMDA signals. Life Sci. 2000; 67: 335-364.

90. Muir GD, Steeves JD. Sensorimotor stimulation to improve locomotor recovery after spinal cord injury. Trends in Neurosciences 1997; 20: 72-77.

91. Popovic MR, Popovic DB, Keller T. Neuroprostheses for grasping. Neurol.Res. 2002; 24: 443-452.

92. Wieler M, Stein RB, Ladouceur M, Whittaker M, Smith AW, Naaman S et al. Multicenter evaluation of electrical stimulation systems for walking. Arch.Phys.Med Rehabil. 1999; 80: 495-500.

93. Agnew WF, McCreery DB, Yuen TG, Bullara LA. Evolution and resolution of stimulation-induced axonal injury in peripheral nerve. Muscle Nerve 1999; 22: 1393-1402.

94. Brummer, S. B. and Turner, M. Electrochemical aspects of neuromuscular stimulation. Functional Neuromuscular Stimulation: Report of a Workshop. 1972. Washington, National Academy of Sciences.

95. Agnew WF, McCreery DB, Yuen TG, Bullara LA. MK-801 protects against neuronal injury induced by electrical stimulation. Neuroscience 1993; 52: 45-53.

96. Deurloo, K. Selectivity in extraneural stimulation of peripheral nerves. 8-21. 1999. Twente University.

97. Popovic MR, Curt A, Keller T, Dietz V. Functional electrical stimulation for grasping and walking: indications and limitations. Spinal Cord. 2001; 39: 403-412.

98. Cochlear implants in adults and children. NIH Consens.Statement 1995; 13: 1-30.

99. Vodovnik L, Kralj A, Stanic U, Acimovic R, Gros N. Recent applications of functional electrical stimulation to stroke patients in Ljubljana . Clin.Orthop. 1978; 64-70.

100. Burridge JH, Taylor PN, Hagan SA, Wood DE, Swain ID. The effects of common peroneal stimulation on the effort and speed of walking: a randomized controlled trial with chronic hemiplegic patients. Clin.Rehabil. 1997; 11: 201-210.

101. Wieler, M., Naaman, S., and Stein, R. B. WalkAid: An improved functional electrical stimulator for correcting foot-drop. 101-104. 1996. Cleveland, Ohio. Proc 1st Ann Conf Int Functional Electrical Stimulation Society.

102. Kenney L, Bultstra G, Buschman R, Taylor P, Mann G, Hermens H et al. An implantable two channel drop foot stimulator: initial clinical results. Artif.Organs 2002; 26: 267-270.

103. Taylor PN, Burridge JH, Dunkerley AL, Wood DE, Norton JA, Singleton C et al. Clinical use of the Odstock dropped foot stimulator: its effect on the speed and effort of walking. Arch.Phys.Med Rehabil. 1999; 80: 1577-1583.

104. Taylor PN, Burridge JH, Dunkerley AL, Lamb A, Wood DE, Norton JA et al. Patients' perceptions of the Odstock Dropped Foot Stimulator (ODFS). Clin.Rehabil. 1999; 13: 439-446.

105. Lyons GM, Sinkjaer T, Burridge JH, Wilcox DJ. A review of portable FES-based neural orthoses for the correction of drop foot. IEEE Trans.Neural Syst.Rehabil.Eng 2002; 10: 260-279.

106. Kralj A, Bajd T, Turk R. Electrical stimulation providing functional use of paraplegic patient muscles. Med Prog.Technol. 1980; 7: 3-9.

107. Bajd T, Kralj A, Turk R, Benko H, Sega J. The use of a four-channel electrical stimulator as an ambulatory aid for paraplegic patients. Phys.Ther. 1983; 63: 1116-1120.

108. Kralj A, Bajd T, Turk R. Enhancement of gait restoration in spinal injured patients by functional electrical stimulation. Clin.Orthop. 1988; 34-43.

109. Brindley GS, Polkey CE, Rushton DN. Electrical splinting of the knee in paraplegia. Paraplegia 1979; 16: 428-437.

110. Donaldson ND, Perkins TA, Worley AC. Lumbar root stimulation for restoring leg function: stimulator and measurement of muscle actions. Artif.Organs 1997; 21: 247-249.

111. Kobetic R, Triolo RJ, Marsolais EB. Muscle selection and walking performance of multichannel FES systems for ambulation in paraplegia. IEEE Trans.Rehabil.Eng 1997; 5: 23-29.

112. Graupe D, Davis R, Kordylewski H, Kohn KH. Ambulation by traumatic T4-12 paraplegics using functional neuromuscular stimulation. Critical Reviews in Neurosurgery 1998; 8: 221-231.

113. Graupe D, Kohn KH. Functional neuromuscular stimulator for short-distance ambulation by certain thoracic-level spinal-cord-injured paraplegics. Surg Neurol. 1998; 50: 202-207.

114. Popovic MR, Keller T, Pappas IP, Dietz V, Morari M. Surface-stimulation technology for grasping and walking neuroprosthesis. IEEE Eng Med Biol.Mag. 2001; 20: 82-93.

115. Scott TR, Peckham PH, Keith MW. Upper extremity neuroprostheses using functional electrical stimulation. Baillieres Clin.Neurol. 1995; 4: 57-75.

116. Snoek GJ, IJzerman MJ, in 't Groen FA, Stoffers TS, Zilvold G. Use of the NESS handmaster to restore handfunction in tetraplegia: clinical experiences in ten patients. Spinal Cord. 2000; 38: 244-249.

117. Prochazka A, Gauthier M, Wieler M, Kenwell Z. The bionic glove: an electrical stimulator garment that provides controlled grasp and hand opening in quadriplegia . Arch.Phys.Med Rehabil. 1997; 78: 608-614.

118. Handa Y, Handa T, Ichie M, Murakami H, Hoshimiya N, Ishikawa S et al. Functional electrical stimulation (FES) systems for restoration of motor function of paralyzed muscles--versatile systems and a portable system. Front Med Biol.Eng 1992; 4: 241-255.

119. Rebersek S, Vodovnik L. Proportionally controlled functional electrical stimulation of hand. Arch.Phys.Med Rehabil. 1973; 54: 378-382.

120. Popovic, D., Popovic, M., Stojanovic, A., Pjanovic, A., Radosavljevic, S., and Vulovic, D. Clinical evaluation of the Belgrade Grasping System. 247-250. 1998. Vienna, Austria. Proc 6th Vienna Internat Workshop on Functional Electrical Stimulation.

121. Keith MW. Neuroprostheses for the upper extremity. Microsurgery 2001; 21: 256-263.

122. Carroll S, Cooper C, Brown D, Sormann G, Flood S, Denison M. Australian experience with the Freehand System for restoring grasp in quadriplegia. Aust.N.Z.J Surg 2000; 70: 563-568.

123. Popovic D, Stojanovic A, Pjanovic A, Radosavljevic S, Popovic M, Jovic S et al. Clinical evaluation of the bionic glove. Arch.Phys.Med Rehabil. 1999; 80: 299-304.

124. Thorsen R, Spadone R, Ferrarin M. A pilot study of myoelectrically controlled FES of upper extremity. IEEE Trans.Neural Syst.Rehabil.Eng 2001; 9: 161-168.

125. Glenn WW, Hogan JF, Loke JS, Ciesielski TE, Phelps ML, Rowedder R. Ventilatory support by pacing of the conditioned diaphragm in quadriplegia. N.Engl.J Med. 1984; 310: 1150-1155.

126. Glenn WW, Phelps ML, Elefteriades JA, Dentz B, Hogan JF. Twenty years of experience in phrenic nerve stimulation to pace the diaphragm. Pacing Clin.Electrophysiol. 1986; 9: 780-784.

127. Brouillette RT, Marzocchi M. Diaphragm pacing: clinical and experimental results. Biol.Neonate 1994; 65: 265-271.

128. Glanz M, Klawansky S, Stason W, Berkey C, Chalmers TC. Functional electrostimulation in poststroke rehabilitation: a meta-analysis of the randomized controlled trials. Arch.Phys.Med Rehabil. 1996; 77: 549-553.

129. de Kroon JR, van der Lee JH, IJzerman MJ, Lankhorst GJ. Therapeutic electrical stimulation to improve motor control and functional abilities of the upper extremity after stroke: a systematic review. Clin.Rehabil. 2002; 16: 350-360.

130. Ada L, Foongchomcheay A. Efficacy of electrical stimulation in preventing or reducing subluxation of the shoulder after stroke: a meta-analysis. Aust.J Physiother. 2002; 48: 257-267.

131. Schmidt RA, Tanagho EA. Feasibility of controlled micturition through electric stimulation. Urol Int. 1979; 34: 199-230.

132. Groen J, Bosch JL. Neuromodulation techniques in the treatment of the overactive bladder. BJU.Int. 2001; 87: 723-731.

133. Hassouna MM, Siegel SW, Nyeholt AA, Elhilali MM, van Kerrebroeck PE, Das AK et al. Sacral neuromodulation in the treatment of urgency-frequency symptoms: a multicenter study on efficacy and safety. J Urol. 2000; 163: 1849-1854.

134. Bosch JL, Groen J. Sacral (S3) segmental nerve stimulation as a treatment for urge incontinence in patients with detrusor instability: results of chronic electrical stimulation using an implantable neural prosthesis. J Urol. 1995; 154: 504-507.

135. Brindley GS, Polkey CE, Rushton DN. Sacral anterior root stimulators for bladder control in paraplegia. Paraplegia 1982; 20: 365-381.

136. Brindley GS. The first 500 patients with sacral anterior root stimulator implants: general description. Paraplegia 1994; 32: 795-805.

137. van der Aa HE, Alleman E, Nene A, Snoek G. Sacral anterior root stimulation for bladder control: clinical results. Arch.Physiol Biochem. 1999; 107: 248-256.

138. Van Kerrebroeck EV, van der Aa HE, Bosch JL, Koldewijn EL, Vorsteveld JH, Debruyne FM. Sacral rhizotomies and electrical bladder stimulation in spinal cord injury. Part I: Clinical and urodynamic analysis. Dutch Study Group on Sacral Anterior Root Stimulation. Eur.Urol 1997; 31: 263-271.

139. Wielink G, Essink-Bot ML, van Kerrebroeck PE, Rutten FF. Sacral rhizotomies and electrical bladder stimulation in spinal cord injury. 2. Cost-effectiveness and quality of life analysis. Dutch Study Group on Sacral Anterior Root Stimulation. Eur.Urol 1997; 31: 441-446.

140. Siegel SW, Catanzaro F, Dijkema HE, Elhilali MM, Fowler CJ, Gajewski JB et al. Long-term results of a multicenter study on sacral nerve stimulation for treatment of urinary urge incontinence, urgency-frequency, and retention. Urology 2000; 56: 87-91.

141. Bosch JL, Groen J. Neuromodulation: urodynamic effects of sacral (S3) spinal nerve stimulation in patients with detrusor instability or detrusor hyperflexia. Behav.Brain Res. 1998; 92: 141-150.

142. Thon WF, Baskin LS, Jonas U, Tanagho EA, Schmidt RA. Neuromodulation of voiding dysfunction and pelvic pain. World J Urol 2003; 9: 138.

143. Elabbady AA, Hassouna MM, Elhilali MM. Neural stimulation for chronic voiding dysfunctions. J Urol. 1994; 152: 2076-2080.

144. Jonas U, Fowler CJ, Chancellor MB, Elhilali MM, Fall M, Gajewski JB et al. Efficacy of sacral nerve stimulation for urinary retention: results 18 months after implantation. J Urol. 2001; 165: 15-19.

145. Vapnek JM, Schmidt RA. Restoration of voiding in chronic urinary retention using neuroprothesis. World J Urol 2003; 9: 142.

146. Marani E. Innervation of the Mature Human Pelvis. The Pelvis: Another View. Enschede: Twente University Press, 2002: 95-112.

147. Dobelle WH, Mladejovsky MG. Phosphenes produced by electrical stimulation of human occipital cortex, and their application to the development of a prosthesis for the blind. J Physiol 1974; 243: 553-576.

148. Bartlett JR, Doty RW. An exploration of the ability of macaques to detect microstimulation of striate cortex. Acta Neurobiol.Exp.(Wars.) 1980; 40: 713-727.

149. Schmidt EM, Bak MJ, Hambrecht FT, Kufta CV, O'Rourke DK, Vallabhanath P. Feasibility of a visual prosthesis for the blind based on intracortical microstimulation of the visual cortex. Brain 1996; 119 ( Pt 2): 507-522.

150. Dobelle WH. Artificial vision for the blind by connecting a television camera to the visual cortex. ASAIO J 2000; 46: 3-9.

151. Veraart C, Raftopoulos C, Mortimer JT, Delbeke J, Pins D, Michaux G et al. Visual sensations produced by optic nerve stimulation using an implanted self-sizing spiral cuff electrode. Brain Res. 1998; 813: 181-186.

152. Grumet AE, Wyatt JL, Jr., Rizzo JF, III. Multi-electrode stimulation and recording in the isolated retina. J Neurosci.Methods 2000; 101: 31-42.

153. Chow AY, Chow VY. Subretinal electrical stimulation of the rabbit retina. Neurosci.Lett. 1997; 225: 13-16.

154. Zrenner E, Miliczek KD, Gabel VP, Graf HG, Guenther E, Haemmerle H et al. The development of subretinal microphotodiodes for replacement of degenerated photoreceptors. Ophthalmic Res. 1997; 29: 269-280.

155. Weiland JD, Humayun MS, Dagnelie G, de Juan E Jr, Greenberg RJ, Iliff NT. Understanding the origin of visual percepts elicited by electrical stimulation of the human retina. Graefes Arch.Clin.Exp.Ophthalmol. 1999; 237: 1007-1013.

156. Rizzo JF, III, Wyatt J, Loewenstein J, Kelly S, Shire D. Methods and perceptual thresholds for short-term electrical stimulation of human retina with microelectrode arrays. Invest Ophthalmol.Vis.Sci. 2003; 44: 5355-5361.

157. Humayun MS, Weiland JD, Fujii GY, Greenberg R, Williamson R, Little J et al. Visual perception in a blind subject with a chronic microelectronic retinal prosthesis. Vision Res. 2003; 43: 2573-2581.

158. Barbeau H, McCrea DA, O'Donovan MJ, Rossignol S, Grill WM, Lemay MA. Tapping into spinal circuits to restore motor function. Brain Res.Brain Res.Rev. 1999; 30: 27-51.

159. Kargo WJ, Giszter SF. Rapid correction of aimed movements by summation of force-field primitives. J Neurosci. 2000; 20: 409-426.

160. Dimitrijevic MR, Gerasimenko Y, Pinter MM. Evidence for a spinal central pattern generator in humans. Ann N.Y.Acad.Sci. 1998; 860: 360-376.

161. Richardson RR, Cerullo LJ, McLone DG, Gutierrez FA, Lewis V. Percutaneous epidural neurostimulation in modulation of paraplegic spasticity. Six case reports. Acta Neurochir.(Wien.) 1979; 49: 235-243.

162. Pinter MM, Gerstenbrand F, Dimitrijevic MR. Epidural electrical stimulation of posterior structures of the human lumbosacral cord: 3. Control Of spasticity. Spinal Cord. 2000; 38: 524-531.

163. Deurloo KE, Holsheimer J, Boom HB. Transverse tripolar stimulation of peripheral nerve: a modelling study of spatial selectivity. Med Biol.Eng Comput. 1998; 36: 66-74.

164. Deurloo KE, Holsheimer J, Bergveld P. Nerve stimulation with a multi-contact cuff electrode: validation of model predictions. Arch.Physiol Biochem. 2000; 108: 349-359.

165. Rutten WL, Smit JP, Frieswijk TA, Bielen JA, Brouwer AL, Buitenweg JR et al. Neuro-electronic interfacing with multielectrode arrays. IEEE Eng Med Biol.Mag. 1999; 18: 47-55.

166. Smit JP, Rutten WL, Boom HB. Endoneural selective stimulating using wire-microelectrode arrays. IEEE Trans.Rehabil.Eng 1999; 7: 399-412.

167. Frieswijk TA, Smit JP, Rutten WL, Boom HB. Force-current relationships in intraneural stimulation: role of extraneural medium and motor fibre clustering. Med Biol.Eng Comput. 1998; 36: 422-430.

168. Zhou BH, Baratta R, Solomonow M. Manipulation of muscle force with various firing rate and recruitment control strategies. IEEE Trans.Biomed.Eng 1987; 34: 128-139.

169. Feirabend HK, Choufoer H, Ploeger S, Holsheimer J, van Gool JD. Morphometry of human superficial dorsal and dorsolateral column fibres: significance to spinal cord stimulation. Brain 2002; 125: 1137-1149.

170. Fang ZP, Mortimer JT. A method to effect physiological recruitment order in electrically activated muscle. IEEE Trans.Biomed.Eng 1991; 38: 175-179.

171. Fang ZP, Mortimer JT. Selective activation of small motor axons by quasi-trapezoidal current pulses. IEEE Trans.Biomed.Eng 1991; 38: 168-174.

172. van Bolhuis AI, Holsheimer J, Savelberg HH. A nerve stimulation method to selectively recruit smaller motor-units in rat skeletal muscle. J Neurosci.Methods 2001; 107: 87-92.

173. Grill WM, Mortimer JT. Inversion of the current-distance relationship by transient depolarization. IEEE Trans.Biomed.Eng 1997; 44: 1-9.

174. Deurloo KE, Holsheimer J, Bergveld P. The effect of subthreshold prepulses on the recruitment order in a nerve trunk analyzed in a simple and a realistic volume conductor model. Biol.Cybern. 2001; 85: 281-291.