Applied Neural Control in the 1990 s

The authors describe some of the current neural prostheses and examine technological developments needed for future generations of neural prosthetic implants. Current developments include the peroneal nerve stimulator, upper and lower extremity functional neuromuscular stimulation and the auditory prosthesis. Three issues connected with future developments include stimulating and recording electrodes, the interconnection system, encapsulation, the command unit, force, touch, and position sensors, and signal conditioning. >

to n e w appl/cat/ons.In th/s paper we descrrbe 50me of the current neural prostheses a n d examine technologlcal developments n e e d e d for future generations of neural prosthet/c /rnplants.I.I I\: I K0I)CJC' 1 ION A healthy, 28-year-old electrical engineer feels some weakness when getting out of bed.H e attributes this t o the softball game he played the night before and proceeds to work.Later i n the day he n o t i w s slight blurred vision and some hesitancy while urinating.He takes the remainder of the day off, rests in bed that evening and the next morning is teeling better.H e does well tor a week, then awakens unable to move either leg and unable to experience sensation below the mid-chest area.H e also notes further progression of his blurred vision.H e i s taken to the hospital where extensive diagnostic procedures are performed.His vision continues to deteriorate for several days then stabilizes.At this time, he has impaired vision, is paralyzed from the waist down, and has n o sensation below the midchest level.
The diagnostic tests contirm the initial impression of the neurologist that he is suffering from Devix syndrome, acondition related t o multiple sclerosis.Some individuals have some recovery i n this disease but others are left with permanent paralysis.Atter his stay i n the acute care hospital he is transferred to a rehabilitation hospital.
At the rehabilitation hospital he i s trained t o use his remaining senses and muscles to substitute for the functionally disconnected ones.The success of this approach to rehabilitation is demonstrated by the round-the-world wheel chair journey of Rick Hansen [40].A more desirable approach, though, is reconnection of the disconnected muscles and sense organs to the central nervous system.Two approaches t o reconnection are central nervous system regeneration and neural prostheses.At this time functional regeneration i s not possible.I n the discussion that follows, w e outline some of the technological problems to M a n u x r i p t received November 9, 1987,  be solved for neural prosthetic systems to benefit significantly patients w i t h neurologic deficits.The damaged nervous system presents special problems and opportunities t o the health care community.A relatively small break i n the system can leave major muscle groups or portions of the brain without input or output.It is a great frustration t o physicians t o k n o w that a healthy muscle cannot be used because there i s n o practical way to stimulate it or that motor areas of the brain, specialized t o perform coordinated tasks, are unusable because they are n o longer connected to their muscles, or that the visual processing centers i n the brain cannot be used because  [55] report that "gadget into1erance"of the external portion of the device was a universal complaint.Thi5 stemmed both from frequent failures of the external portion of the unit and from difficulties i n d o n n i n g it b y a hemiplegic individual o n a daily basis.Also, using a single rhannel device, it has been difficult to obtain balanced flexion of the ankle.There is a tendency for the stimulated ankle to twist inward (vargus) o r twist outwared (valgus) along w i t h the flexion.This balance problem can be largely solved using t w o channels of stimulation.Gadget intolerance might be solved by developing a fully implanted system or a system that is easier to use.Waters et al. report that they are proceeding t o develop a totally implanted multichannel system for foot drop.This example demonstrates the costibenefit balance that must be shifted t o the benefit side if neural prosthetic systems are t o aid patients.I n this case, the benefit will be improved by going t o a t w o channel system and the cost will be reduced bv dcvcloping a more reliable, tully implantable systern which removes the c o s t ot daily der.-Tht, technology ot stimulation t o maintain muscle mass is well understood 1381.Walking systems are undergoing feasibililty studies and clinical development.Walking i n a normal individual is a complex process that makes use of visual, vestibular, tactile, and proprioceptive senses as well as motor responses.Walking systems for paraplegic individuals that successfully compete w i t h a wheel(-hair will require technologic advances i n force and position sensors, inertial reference systems, and fast, closed loop controllers that can stimulate reflexes t o prevent falls as well as advances i n electrodesand stimulatorsespecially ifwalking over uneven terrain is required.

D. The Auditory Prosthesis
The cochlear prosthesis transforms sounds picked u p with a microphone into electrical signals that are used to generate auditory sensations i n deaf individuals by stimulating their cochlear nerve.A recent review of these systems was published by Loeb [33].A sensory prosthesis ot this type is i n many ways simpler than a motor prosthesis.There are n o control problems.The electrode and implanted electronics are i n an environment where there is essentially n o relative motion.Muscle fatigue is not present.It is not surprising then that the clinical dcvclopment of these systems has proceeded must faster than the development of motor systems.Reliable implants are commercially available at this time 1211, 1241, [48].
All auditory prostheses, whether single c-hannel or multiplechannel, improve lip reading and provide the user with valuable auditory cues about what is occurring i n the environment [5].Even with single channel devices some patients are able to recognize speech without visual cues, but they are the exception [22].
Improvements i n the signal processing o i spcec-h before it is presented to the stimulating elec-trodes may permit more patients t o recognize speech.At this time, n o single speech processing technique has proven superior for all patients [57].However, selected patients d o significantly better using one speech processing approach or another.The reasons for this variability need to be determined.
Little is k n o w n about the value of auditory prostheses i n young children w h o are learning speech and language.Although over250children have received cochlear implants 1251, these have been simple, singlechannel devices.Betore more complex multichannel systems can be recommended for widespread evaluation more information is needed about the effects of head growth o n the implant [39] and about means of preventing spread of otitis media intections along the implant hardware which is likely to occur i n this age population.Solutions t o these potential problems are n o w being studied.
Cochlear implants are of n o value to individuals w h o have total degeneration of the auditory nerves.There is also evidence that implant subjects w h o have very limited nerve survival generally d o p o o r l y w i t h acochlear implant 1-59], For these groups, an auditory prosthesis which interfaces with the auditory nervous system more centrally may be of value.House and colleagues have implanted electrodes i n several patients whose auditory nerves had been resected during surgical removal of bilateral acoustic neuromas [36].These electrodes consisted of a pair of bipolar platinum plates placed over the cochlear nucleus i n the brainstem.The results have been similar t o single channel scala tympani implants (cochlear implants) i n patients with good nerve survival.It is likely that an array of microelectrodes implanted into the cochlear nucleus and controlled by a multichannel speech processor will provide more useful iniormation than a pair o t electrodes placed o n thecochlear nucleus surface.However before this is attempted, ieasibility iniormation is nceded trom animal studies t o answer questions about the best surgical approach and whether the cochlear nucleus can be saiely stimulated for long periods of time.I i the results of these animal studies are positive, studies ot microstimulation o f t h e human cochlear nucleus i n deat individuals will surely oc.cur.Currents as l o w as ten t o twenty microamperes w ,I1 stimulate small populations of CNS cells when microelectrodes are used.
Electrodes provide the interface between the electronic current i n the lead wires and the ionic current i n the tissue fluid.They must safely and effectively provide the necessary stimulating current.Recognizing that there is a thousand-fold range i n the stimulating current for different electrode locations, it is not surprising that the elec-trodes for each application have special problems.
Stainless steel and platinum electrodes provide adequate charge transfer per u n i t area w h e n relatively large electrodes are used as i n stimulating muscle.W i t h smaller electrodes, the electrode surface area decreases more rapidly than the current requirements and as a result, the electrode-tissue current density and charge density transfer requirements for microstimulation are greater than for stimulation using larger electrodes.Pulse current densities of u p to fifty amperes per square centimeter are needed for some electrodes w i t h small surface areas.Approaches t o developing electrodes with high charge capacity per unit area include increasing their effective surface area and increasing their charge capacity.Currently the leading material for high charge transfer is "activated iridium" which can deliver significantly higher charge densities than platinum [q.New electrode materials and electrode designs will besoughtandevaluated.M e t h o d s t o increasetheeffective surface area will be explored.I n addition, the optimal electrode size for microstimulation needs t o be established by experimentation i n animals and eventually i n man.A variety of electrodes will be developed for specific applications.These include intramuscular electrodes, epimysia1 electrodes [18], nerve cuff electrodes [53], flexible electrode arrays, and rigid electrode arrays.

Recording Electrodes
Neurons haveassociated extracellular potentials that can be recorded using small electrodes placed near an excited neuron.These potentials are candidates for control signals i n future motor prostheses.Themajor problemstoconquer with recording electrodes are improved selectivity, prevention of motion between the electrode and the recorded neuron, and minimization of mechanical damage to the neural tissue.The charge transfer at the electrode is small relative t o stimulating electrodes so high current densities are not a factor.High effective surface areas are desirable to minimize electrode thermal noise.Mechanical design advances that minimize mechanical injury at the implant site are needed.Recording techniques under development include nerve cuff electrodes [23], electrodes within nerves [15], integrated circuit electrodes [2], [27], [28] and the thumbtack electrode [46].The evolution of these recording electrodes i s likely t o continue, perhaps requiring only minor geometric changes o r more judicious implant site selection to provide marked improvement i n long term recording ability.The addition of amplifiers and multiplexer circuitry t o the electrode arrays will increase the number of available channels while reducing the number of required interconnecting leads.Improvements i n the interconnection systems can also be expected to increase the survival time of chronic recording electrodes.
trolled and powered by a radio frequency electromagnetic link.Such a micro-sized device should be small enough t o allow implantation through a hypodermic needle.

D. Encapsulation C. The Interconnection System
Long tract neurons are a major element of the nervous system.They function t o connect sensors (touch, position, etc.) and motors (muscles, blood vessel walls, sweat glands, etc.) with the central nervous system.I n a similar way, prosthetic sensors and electrodes must be interconnected i n a neural prosthetic system.I n addition, power is usually required at the sensors and electrodes.For simple systems with only a few channels, insulated wires provide the obvious method for interconnection.This approach has been quite successful and is used in thecardiac pacemaker, the cochlear prosthesis, and in F N S systems rurrently i n use.Significant effort has been directed over the past several years t o making reliable lead systems.Despite these efforts, leads continue t o fail [51], [52].Problems include stress fatigue failure of the lead wires as they pass through tissue planes and corrosion failure i n association w i t h insulation failure.Marsolais [34] reports a failure rate of about four percent per month for percutaneous electrodes that survived over six months i n his FNS application.Wires that are subjected to less movement such as i n the cochlear prosthesis have much better survival.However, even this situation presents significant risk ot failure as a result of corrosion when the lead wires are subjected t o cathodal o r anodal potentials w i t h respect t o the extracellular fluid.Providing an interconnection system for electrodes placed i n a moving muscle is fundamentally different than for electrodes within the relatively stable CNS.It is interesting to note that the body also recognizes that there are different environments for its o w n interconnects by providing the long tract nerves outside the CNS w i t h protective connective tissue sheaths which are not needed by CNS tracts.
For prostheses that interface w i t h the CNS, multiconductor microcables connecting directly w i t h the bonding pads of integrated circuit electrodes are needed.These cables will encounter relatively little mechanical stress but will be subjected t o corrosive potentials.Further research and development o n insulating polymers are needed to provide microcables w i t h lifetimes that match human lifetimes.Also, studies of the relation between corrosion and the carrier frequency used for carrying power and control signals must be examined.Integrated circuit electrodes might incorporate a r t o dc converters if cables carrying ac are more corrosion resistant.The lead wires for muscle electrodes and peripheral sensors need t o be more resistant t o mechanical stress such as that caused by repeated flexing as they pass over joints.Conductive polymers may prove useful i n such cables.
Efforts are needed t o develop more reliable cables.Equally important, efforts are needed t o develop alternatives t o the milticonductor cable as an interconnect.This becomes especially critical i n applications such as a visual prosthesis where large numbers of stimulating electrodes will be required.Alternatives t o a cable for interconnection include coupling by ultrasound, radio frequency induction, and light.O n e possible system might include numerous micro-sized receiver-stimulator-electrode units implanted at the desired sites of stimulation which are con-Neural prosthetic implants must be compatible with the body.The body must be protec-ted trom the implant and the implant from the body.Protecting the implant is a significant problem.The pacemaker industry has developed effective encapsulation using the hermetically sealed titanium can.Integrated circuit electrodes and sensors require encapsulation that is less bulky.
Multichannel systems will use electrodes and sensors with electronics integrated into their structures.T h r y will have interconnects which may be discrete wires, ribbon cables, or wireless links.Protection of these integrated structures is difficult because their micron-width conduc-tor paths are at potentials of several volts relative t o each other and relative to the body extracellular iluid.Advanced methods of encapsulation will be required.Hermetic surface passivation may be adequate using glasses such as silicon dioxide and silicon nitride perhaps i n association with metal films.Some implants might be fully encased i n an isolating capsule, but others like micro-sized receiver-stimulator-electrodes must have encapsulant edgcs that allow exposure of the electrode t o the biological environment.
It is apparent that several approaches t o encapsulation must be developed for speciiic neural prosthetic applic-ations.The barrier approach involves coating the implant with a material that is essentially irnpermeablc.t o water and dissolved ions.Thegoal i s a t h i n f i l m e q u i v a l e n t o t hermetic encapsulation.This may be appropriate tor protection of planar integrated circuit structures.Another approach, the selective barrier, w o u l d use a polymer coating t o protect the integrated circuits and lead wires i n the extracellular fluid environment.The coating need not bc impermeable t o a l l molecules b u t o n l y t o t h o s e t h a t tend t o makctheelectrodes tail such as sodium ions.Tethering cables used t o interconnect electrodes and sensors could beneiit i r o m corrosion protection using flexible, tightly adhering, thin, insulating films.These d o not have t o be impermeable to water vapor.

E. The Command Unit
Some FNS systems such as the cardiac pacemakt,r and the peroneal nerve stimulator require littleor novolitional control.Others suc-h as FNS systems for restoring hand function, standing, or walking require consc.iou5control t o initiate and coordinate movements.Control (onimands t a n be obtained from movements of unparalyzcd muscles.For example, paraplegic individuals can operate joy sticks and switches with their hands to provide commands.Quadriplegic individuals can move their shoulders, head, ne( k, eyesand tongue.Voice provides another potential channel, but it draws attention t o a person's disability and intertercs with speech.The shoulder movement system being drveloped at Case Western Reserve University provides proportional and on-off commands along orthogonal axes o i the shoulder [44].Fully implanted systems that detect movement o f joints or muscles need t o be developed.Low power external or intraoral control channels that communicate through a telemetry link also need t o be developed.
It is desirable t o derive commands from those portions of the central nervous system that are still functional but are n o longerconnected t o a motoroutput.Thiswil1 require the ability t o record from cells i n the motor cortex or other areas proximal to the lesion [47].The integrated circuit recording electrode allows recording from small populations of cells.Final fabrication and encapsulation of the electrode, development of an interconnection system, and demonstration of voluntary control over the activity ot selected small populations of cells i n the cortex of human volunteers are needed t o demonstrate the feasibility of this approach.
F. Force, Touch, and Position Sensors Operational sensors for force, touch, position, and temperature are present i n most paralyzed individuals but are disconnected from the brain.Neural prosthetic systems might use electrodes to derive signals directly from these sensors.An alternate approach might use artificial sensors and bypass the biological sensors altogether.Both approaches have promise, and it seems likely that both will be refined and find application i n tuture neural prosthetic systems.
The use of intact biological sensors requires the ability to record chronically from small populations of cells i n a peripheral nerve or i n the CNS.Significant signal processing will be required t o transform the recorded action potentials into measures of force, touch, and position.Loeb [30] recorded from sensory fibers i n the dorsal root ganglia of the unrestrained cat tor periods of t w o months.Hoffer [23] has demonstrated recordings ot neural signals encoding force using a cuff electrode o n the tibial nerve in cats.The touch receptors are served by cutaneous nerve branches that might be tapped t o give a moderate degree of selectivity.Significant electrode design and interconnection problems must be overcome to make the peripheral electrode practical.Operation over dec-ades requires that the nerve must survive with the electrode i n or around it.This i n t u r n requires a highlyflexible electrodeand interconnect or a means of mechanically stabilizing the nerve.Placing an electrode morecentrally such as i n thedorsal roots solves some of the movement problems associated with a peripheral electrode.However, the surgery becomes more complex, and the risk of CNS infection increases.
I n the short term, artificial transducers may provide a better source of sensory input.The heel switch used to trigger the stimulus for the foot d r o p prosthesis is one of the simplest sensory transducers.An external goniometer (joint angle transducer) t o monitor angle i n the finger joints is being developed at Case Western Reserve University.Arrays of pressure transducers are also under development [ I l l .These might ultimately be implantable or be put o n like a glove.An implantable goniometer has been developed at Illinois Institute of Technology that uses the Hall effect for the transducer mechanism [54].The interconnection problem remains whether biological or artificial transducers are used to provide sensory information.ual's sensory nervous system.I n practice, though, some ot the sensory substitution is likely t o be done using artiiic ial transducers.

G. Control Systems
The stimulus response relationship i n muscle is noiiliiiear and nonstationary, and muscles can show rapid fatigue.Closed loop control can compensate for these charac-teristics and also improve the efficiencyof a prosthetic system [lO].Astiffness regulation feedbackcontrol system for hand grasp i n quadriplegic individuals is undergoing laboratory evaluation [12].The control approaches t o solving problems such as fatigue i n muscles will depend heavily o n the number of channelsavailablefor stimulation and hence, o n the number of different ways of producing the desired output.The complexity of the control systems will be directly related to the different functions to beaccomplished which will i n t u r n depend o n the needs of the user, the number of available channels and the sophistication of the user's control unit.For example, walking across a level, hard floor is a different problem than walking u p a rocky incline.The degree of control will also be a function of the quality o f the force, position and slip feedback provided by sensors.

H. Signal Conditioning
In an auditory prosthesis the speech signal from a microphone i s electronically processed before presenting it as an electrical stimulus i n the scala tympani o r i n the cochlear nucleus.An idealistic approach t o this processing is to transform speech into a series of impulses that mimics the physiological response that w o u l d be seen i n the neurons at the stimulation site under conditions of natural hearing.A practical approach is t o generate such a neural response as closely as possible within the constraints of the stimulation system.Another approach extracts key speech features and presents them at different electrical stimulation sites.Psychophysical testing can be used t o determine perceptual channels that are defined i n terms of electric.alstimulation sites [9], [56] [58].Processed speech is then applied to these perceptual channels.At this time some patients d o better with one speech processing technique and others d o better w i t h another.The speech processor represents just one of many potential future signal conditioning svstems that will be required t o transform the scnsory environment into a spatio-temporal pattern of electrical stimulation that is easily interpreted by the user.

I Microstimulation in the Central Nervous System
The microstimulation electrodes presently under drvel opment may make new prostheses b a w d o i l central ncrvous system (CNS) stimulation feasible To exploit thi5 tec hnology, CNS microstimulation studies must be conducted to obtain more detailed understanding ot the neurophysiology and neuroanatomy of motor areas i n the CNS asiociated w i t h specific functions such as micturition, incon tinence, and penile erection.

V. SUMMARY AND CONCL~JSIONS
Neural prosthetic systems are technologically limited at this time.W e can define a successful prosthetic system as one that is used a n d appreciated by t h e handicapped population for whom it is designed.Using this definition, the cardiac pacemaker, the diaphragm pacer, and the cochlear prosthesis are successful.Upper and lower extremity FNS systemswill likely become successful as they become m o r e reliable and fully implanted.The single channel cochlear prosthesis has demonstrated quite convincingly that a limited amount of information suppled to an input starved auditory system is of significant benefit t o deaf individuals.Supplying additional information using multiple channels provides still m o r e useful information.Major goals for current prosthetic systems are t o make t h e m m o r e reliable, m o r e user friendly, and perhaps user invisible like t h e cardiac pacemaker.
Another generation of implants will b e developed that will provide many m o r e channels than current systems.The full impact of these systems is difficult to predict.W e anticipate fatigue resistant FNS systems controlled effortlessly through m o t o r cortex channels f o r t h e upper and lower extremities and for respiration and bladder control.Sensoryprostheses w i l l includeauditory prosthesesforthedeaf that permit visually unaided recognition of speech and visual prostheses f o r t h e b l i n d that permit some rudimentary reading and enhanced mobility.
Neurologic disabilities leave individuals with tremendous untapped resources within their bodies.Medical a n d bioengineering developments will, over time, provide a means of tapping these resources.
revised M a r c h 28,1988 Theauthor5 are w i t h t h e Neural Prosthesis Program, Division of Fundamental Neurosciences, National Institute of Neurological and Communicative Disorders a n d t h e 5troke, National Institutes of Health, Bethesda, MD 20892, U4A IEEE Log N u m b e r 8823240 there is n o way t o provide a substitute stimulus for the lost visual input from the eye.Ideally, neural prosthetic systems w i I I revi tal ize these f u n c t i o n al but d i sco n n ec ted portions of the sensory and motor systems.To d o this, reliable systems must bedeveloped to stimulateand record from small populations of neurons and muscle fibers.Many of the problems t o be solved are engineering ones.Where are we n o w and what are the elements that are needed for future developments?11.CUKRENT DFVELOPMLNT5 IN NEURAL P K O S I H t \ l l A. The Peroneal Nerve Stimulator Functional neuromuscular stimulation (FNS) of the peroneal nerve to correct foot d r o p i n strokc patients was developed over 25 years ago.Yet, it never became widely accepted i n the United States.I n a recent article, Waters et al.

ningH
Upper txtrem/t) F A 5 Stimulation ot the must I f 3 in the torearm and hand i s being used in an FN5 skstern t o restore hand grasp Peckharn 1441 has recently reviewed this tield An individual with a transection ot the spinal cord at the level of the fitth or sixth ( e r \ i ( a l (neck) vertebra usually i \ able t o move his shoulders and tlex his elbows H e might have voluntary niovemcnt ot the wrist However, he will have essentially noability t o m o v e h i s t i n g e r s o r t h u m b a n d thuscannot pick u p o t hold objects F N S ot the upper extremity aims t o restore tunction t o thc paralyLed muscles of the hand Numerous studies using percutaneous electrodes have demonstrated the teasibility ot F N 5 tor providing grasp i n an open loop tashion i n selected patients The individual with even a primati\el\ ( o n t r o l l e d grasp i s able t o d o many routinc dailV c hores su( h as cating, grooming, and even selt (athctcriration that would otherwise require an attendant Although earl\ work has demonstrated the feasibility ot these I N 5 systenis,developmerlt work i s needed t o bring thesetec hniques into routinec Iinical use Fspeciallyneeded are electrodes and stirnulation svstems that will operate reliablv tor the lite ot the uscr A step toward this goal was the implantdtion ot an eight c hdnnel receiver-stimulator svstcrri tor FN5 ot the hand with radio trequency (RF) coupling o t power and control signals 1441, [SO] Grasp i s a tunction that requires voluntary control To allow a uscr t o control his FNS system, Peckham and his (olleagues at Case Westrrn Keserve University mount an external transducer o n the shoulder opposite t o the hand t o b e controlled They have demonstrated that an external joint rnovcincnt transducer can bc1 used effectively to control grasp 1441 I n the tuture, an implanted joint transducer might bc developed based o n angle transducers The precision and spred ot control ot grasp will depend i n part o n the number of degrees ot freedom and the resolution of these transducers and o n the patients' residual motor control Even simple grasp patterns can be ot significant benefit t o the quadriplegic individual O n e simple but elegant approach proposed by Keith (personal commu n icat ion) would useatransduter implanted i n thewrist jointtodetect wrist extension The extension signal w o u l d control the stimulation of the tinger tlexor muscles while grasping an object Thus the normal closure ot the hand, initiated passively by extending the wrist, can be augmented w i t h FNS t o provide a stronger grasp A selection of different grasp modes will require a more sophisticated control channel C. Lower txtremity FNS Lower extremity FNS w a s re( ently reviewed by Cybulski et el.[13].I ower extremity FNS can be used i n applications rangingfrom elec-trical stimulation to maintain muscle mass t o system\ for standing and walking under controlled conditions.

Iv
THt RLLFNI K A I ION APPKOACH Developments should include external and implantable goniometers, implantable force transducers that measure force i n tendons, and touch receptors that can signal t o ntact and slip.I n principle all of these sensors can be implemented by making appropriate connections t o an individ-This discussion has centered on the development ot neural prostheses Regeneration or reconnection acrois lesions is another mechanism whereby neurologic allv impaired individuals might regain function.Someday these techniques might make neural prostheses unnecessary in rehabilitation.Prosthetic systems might then b e used to enhance performance.I n any event, at this time neural prostheses are t h e most practical alternative for the neurologically handicapped.
s undcr development today tor tunctional neuromusc ular stimulation, auditor\/ prostheses, and visual prostheses are likely to undergo signititant improvements over the next several years I n addition, developments are likely i n a new generation ot neural prosthctic syitems that make more exten\ive use ot integratcd electronics at the sensor or electrode level The neural prosthetic systems in use or i n clinical development todav have anywhere trom one to a tew dozen channels This situation i s analogous to the early development ot m i n i and micro computers Because memory was expensive, c oniiderable programmingeftortwasdirectedat achievingthe maximurn usetrom every available byte of memory As the cost ot memory dropped, the emphais i n computer programming 5hitted and currently little consideration i s given t o conserving memory.I n neural prosthetic systems being developed today, considerable ettort i i directcd at making maximal use o t every available channcl It the c o s t per (hannel (in terms ot ccmplications and untoward side ettects, user attention, adjustment and maintenance, space o c ( upied, dollar c o s t , surgical complexity, and s) stem unreliabilityl can be dramatically reduc d, the approach t o current prostheses will shitt away trom channel cltilization toward improved intertates with the user Technological goals in( lude reducing the cost per chan-Short cathodal current pulses ot a tew hundred microsecond5 duration can provide this depolarization Currents ot titteen milliamperes are typically used t o activate motor units i n muscle using electrodes i n close proximity t o the muscle.If electrodes are placed i n or around peripheral nerves, currents of a few hundred microamperes are required to stimulate the nerve' fibers.
[2] by one or t w o orders o t magnitude while simultaneously increasing their reliabilitv This will require even greater reliability ot the indi\ idual elements.To achieve these goals, multichannel svstc'nis will make extensive use ot active integrated circuit elec trodes and iensors[2]I n addition, advances i n interconncct cable technology, nonhermetic encapsulation, and wircleis power and commu n ic at ion system> tor powcri ng t h in-fi I m her met ically encapsu lated re( civer-st i ni u lator-elect rodes and sensors are needed The c,lcments ot neural prosthetic