JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 18(2):25-33, 2008©
Electronic Stimulators for Surface Neural Prosthesis Barry J. Broderick, Paul P. Breen and Gearóid ÓLaighin Abstract—This paper presents the technological advancements in neural prosthesis devices using Functional Electrical Stimulation (FES). FES refers to the restoration of motor functions lost due to spinal cord injury (SCI), stroke, head injury, or diseases such as Cerebral Palsy or Multiple Sclerosis by eliciting muscular contractions through the use of a neuromuscular electrical stimulator device. The field has developed considerably since its inception, with the miniaturisation of circuity, the development of programmable and adaptable stimulators and the enhancement of sensors used to trigger the application of stimulation to suit a variety of FES applications. This paper discusses general FES system design requirements in the context of existing commercial and research FES devices, focusing on surface stimulators for the upper and lower limbs. These devices have demonstrated feasible standing and stepping in a clinical setting with paraplegic patients, improvements in dropped foot syndrome with hemiplegic patients and aided in the restoration of grasping function in patients with upper limb motor dysfunction. Index Terms—Functional Electrical Stimulation, medical devices, neuroprostheses, rehabilitation engineering.
I. INTRODUCTION
N
euroMuscular Electrical Stimulation (NMES) is defined as the application of electrical current to motor points in the body to artificially contract skeletal muscles. Functional Electrical Stimulation (FES) is further defined as the application of NMES in order to facilitate purposeful tasks such as grasping a cup or assisting with walking, in situations where normal neural function has been damaged or destroyed. Thus, FES can be considered as neural prosthesis [1]. The concept was first proposed by Liberson et al. [2] in 1961, who used electrical stimulation applied to the peroneal nerve of hemiplegic patients suffering from drop foot. This elicited a dorsiflexion of the foot synchronised with the swing phase of gait to lift the foot and
prevent it from dragging on the ground during swing. Since 1961, FES has developed considerably and has seen many major technological advances with the implementation of FES devices that allow improved walking in stroke patients [3] and the restoration of upper and lower limb functions in spinal cord injured patients [4-7]. Depending on the intervention, FES may be applied using surface, percutaneous or implanted electrodes. Surface electrodes are the simplest and least invasive implementation and are very useful for a variety of therapies. Two self-adhesive surface electrodes (one anode and one cathode) are placed directly over the motor points of the muscle. Electrode size and placement must be carefully considered in surface stimulation applications as improper placement can directly affect patient comfort levels and muscle contraction strength [8]. With percutaneous FES, a barbed electrode is positioned close to a motor nerve with the aid of a hollow needle with the wire for the electrode passing through the skin [1]. Implanted cuff electrodes are fixed permanently to nerves and are more suitable for long term use. As well as having the electrodes implanted, the lead wires and the stimulator circuitry itself are also typically implanted as a combined unit [7]. The most effective approach to restoring motor functions is to begin FES as soon as possible after injury using a surface system before migrating to an implanted device [6, 9]. Thus, the need for surface FES
Manuscript received November 5th, 2008. This work was supported in part by the Irish Research Council for Science, Engineering and Technology. B. J. Broderick is with the Department of Electrical & Electronic Engineering, National University of Ireland, Galway, Galway, Ireland and the National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Galway, Ireland. P. P. Breen is with the Department of Electrical & Electronic Engineering, National University of Ireland, Galway, Galway, Ireland and the National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Galway, Ireland, phone +353 91 493126, (
[email protected]). G. ÓLaighin is with the Department of Electrical & Electronic Engineering, National University of Ireland, Galway, Galway, Ireland and the National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Galway, Ireland.
DOI:10.2298/JAC0802025B
Fig. 1. Functional electrical stimulator design block diagram.
26
BRODERICK B.J., BREEN P.P., ÓLAIGHIN G.: ELECTRONIC STIMULATORS FOR SURFACE NEURAL PROSTHESIS
devices will remain despite the availability of implanted systems. This paper will focus solely on the design of programmable neural prosthesis based on surface electrical stimulation.
II. FES DESIGN REQUIREMENTS For FES systems to provide therapeutic and orthotic support to patients they must meet certain specific design requirements. To function as a take-home device, they must be safe, portable and be sufficiently easy to operate so that patients can don the equipment unaided or with minimal assistance. The interface must be as ergonomic as possible to allow patients with reduced dexterity or poor eyesight to use the device without difficulty. Most surface FES systems follow a standard electronic design structure, similar to that originally proposed by Ilic et al. [10], consisting of digital control, sensor interface, stimulator, user interface and power supply blocks (Fig. 1). Devices have shifted away from hard-wired designs, such as the device proposed by Liberson [2] where the control of the stimulator is determined by the wiring of its electronic circuitry [3], to microcontroller based devices. Hard-wired designs typically require complicated analog or “dial” based interfaces which make it difficult to apply novel algorithms or fine-tune advanced settings for a specific user. Amplitude Control User Interface +VHIGH Microcontroller Memory
DAC Timing
Level Control (Voltage or current controlled)
High Voltage Converter -VHIGH
Programmer +VHIGH (ADJ)
-VHIGH (ADJ)
Switch Circuitry
The output of surface FES devices can be either constantvoltage, constant-current or a hybrid form of output. The advantage of the constant-voltage setup is that current density determines the potential for tissue damage. As the impedance of the skin increases, current decreases. However, constant-voltage stimulators have a variable motor response. Constant-current stimulators have better contraction consistency and repeatability with less variability in resistance [7]. The strength of the resultant muscular contraction can be determined by varying the stimulus amplitude, pulse width or pulse frequency. In most applications a fused muscle contraction is desirable. To achieve this, stimulation frequencies of up to 50Hz are recommended [11]. Above these stimulation frequencies, fatigue due to non-physiological stimulation rate may be elicited. It is believed that this phenomenon is avoided in normal physiological use as motor neuron firing rates remain below frequencies associated with fatigue [12]. The output pulse train is normally monophasic, asymmetric biphasic or symmetric biphasic (Fig. 3). Monophasic pulses have the potential disadvantage of causing electrode deterioration and tissue damage. As they are capable of altering ionic distributions and causing polarization they can lead to skin breakdown and burns. Asymmetrical biphasic pulse trains are bidirectional and allow ions to flow in both directions, minimizing ion redistribution and subsequent risk of skin irritation. Symmetric biphasic pulses allow both pulses to depolarise the neural membrane. The effect of many chemical reactions are reduced with these pulses [11]. However, the anodic current reversal of a biphasic stimulus can act to abolish an action potential developing in response to the cathodic phase. A delay between pulse phases (interpulse interval) was found to effectively prevent this event from occurring [12]. It has also been reported that the peak voltage required to reach sensory and motor nerve thresholds was lower with biphasic than monophasic stimulation [13]. (a)
Stimulation Electrodes
Fig. 2. Typical embodiment of electric stimulator output block.
(b)
A. Stimulator Block The analog stimulation block of FES systems supplies the high voltages required to elicit muscular contractions and can adjust the output according to the functional requirement. A typical embodiment is illustrated in Fig. 2. A DC-DC converter is used to generate a constant positive and negative high voltage (+VHIGH, -VHIGH). The amplitude/voltage output is then modulated by a digital-toanalog converter (DAC) generating adjusted voltage outputs (+VHIGH(Adj), -VHIGH(Adj)). The DAC level is the result of user input and a waveform shape stored in preprogrammed memory. Timing signals control switch circuitry to generate output pulses of the desired frequency and duration. The required stimulus output is then applied across the stimulation electrodes. The digital circuitry generating the timing signals may need to be isolated from the analog output circuitry due to the large difference in voltage associated with both circuits. This can be accomplished through the use of optocouplers [10].
(c)
(d)
Fig. 3. Common stimulus output trains: (a) Monophasic (b) Asymmetric biphasic (c) Symmetric biphasic (d) Symmetric biphasic with interpulse interval.
Dropped foot stimulators are usually only required to elicit muscular contractions in the tibialis anterior muscle, hence, only one channel of stimulation is required. In cases of more complex motor dysfunction in the lower extremity, 4 or 6 channels of stimulation provide the necessary control of the lower limbs [4]. A novel feature of the Compex
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 18(2): 25-33, 2008©
27
TABLE I THE ANALOG OUTPUT STAGE OF VARIOUS FES DEVICES Features
Bionic Glove
Output Stage
constant-current
Stimulation channels
3
Modulation
Parastep 1
STIWELL med4
Compex Motion
constant-current
constant-voltage
constant-current
constant-voltage
constant-current
Constant-current
4 or 6
4
4
2
8
4
Amplitude
Amplitude
Amplitude/ Pulse Width
Amplitude
Amplitude/ Pulse width
Pulse Train
Biphasic
Monophasic or Biphasic
Biphasic
Monophasic or Biphasic
Biphasic
Monophasic
Biphasic
Pulse Amplitude
25-35V
Up to 70V
100mA 80V (1kΩ)
0-120mA
Up to 72V
0-50mA (150V max)
0-120mA
Pulse Width
100-200µs
120 or 150µs
50µs-250ms
0-16ms
100-500µs
0-1ms
450µs max
Stimulation Frequency
20-30 pulses/sec
24 pulses/sec
0.1-140Hz
1-100Hz
5-50Hz
5-80Hz
120Hz max
Motion system, shown in Fig. 4, is that two or more stimulator units can be cascaded using the special purpose port to allow the synchronised use of additional stimulation channels and meet FES requirements [14]. There is, however, a balance between adding more functionality or stimulation channels to a device to provide further restoration, while maintaining practicality. The more channels required, the more complicated the device becomes for the user to don and the less suitable it will be for take-home use. Therefore, this review has not identified a suitable take-home device where more than 4 channels of stimulation are used. Of course, multichannel stimulation can be used in a clinical setting as an initial motor relearning training period prior to rehabilitation with a smaller 2 or 3 channel take-home device. Table 1 lists the typical electrical characteristics of the output stages of various FES devices.
Duo-STIM
UNAFET 8
CEFAR REHAB X4
Frequency/ Pulse Width
functionality can be added by reprogramming the core without having to modify hardware. Devices such as the UNAFET 8 (UNA Systems, Belgrade, Serbia), the NESS H200 and the NESS L300 (Bioness, Inc., CA, USA), the WalkAide (NeuroMotion, Inc., TX, USA), the Parastep 1 [4], Compex Motion [14] and the Duo-STIM (Fig. 5) [15] are all examples of microprocessor controlled devices. Various parameters can influence the choice of microcontroller used, such as size of memory, built-in timer, DACs and analog-to-digital converters resolution, speed of the core clock and cost. As the functionality of FES devices becomes ever increasingly complex, multiple cores may be required, whereby one core could provide stimulation control and another handle separate tasks such as sensor input sampling [16]. Examples of common microcontrollers that have been used in FES devices are the Hitachi HD637B01VOP [4], Microchip Technology's PIC16F84 [17], the Motorola 68HC11 [10, 14] and the Analog Devices ADuC831 [15]. The digital control block allows execution of sophisticated stimulation algorithms such as an adaptive
Fig. 4. The Compex Motion FES system showing the stimulator itself, electrodes, EMG sensors and some data cards.
B. Digital Control Block A digital control block consisting of a microprocessor and an interfaced memory module offers precise control over stimulation algorithms and waveforms. The processor acts as a dedicated stimulation controller so that specially designed stimulation profiles can be implemented and evaluated. The microcontroller typically generates two sets of control signals according to the selected algorithm, one to control the output pulse shape and a second to control the output stimulus amplitude. The microcontroller can store user parameters in memory and can be interfaced easily with external personal computers (PC) or controllers. Further
Fig. 5. The Duo-STIM programmer unit (a) and stimulator unit (b) (Reprinted with permission).
intensity envelope investigated by Breen et al. for drop foot correction (Fig. 6). Two foot switches, one placed under the heel and one placed under the toe, monitor the patient's gait cycle. The micro-controller then predicts the current gait cycle time based on the previous gait cycle time and dynamically adjusts the stimulation output current of a drop
28
BRODERICK B.J., BREEN P.P., ÓLAIGHIN G.: ELECTRONIC STIMULATORS FOR SURFACE NEURAL PROSTHESIS
Fig. 6. (Left) Adjustment of linear stimulus envelope if actual event-time is later than expected. Stimulation intensity level is maintained until the expected event occurs. Stimulation envelope is now shifted forward in time following this event and the envelope’s rate of change with time is reduced to reflect the longer than expected stride time. (Right) Adjustment of linear stimulus envelope if actual event-time is earlier than expected. The preset stimulation intensity corresponding to that event is immediately delivered. The stimulation envelope is shifted backward in time and the envelope’s rate of change with time is increased to reflect the shorter than expected stride time [18] (Reprinted with permission).
foot stimulator to reflect changes in a patient's stride time [18]. Hart et al. also took advantage of the microprocessor's flexibility by implementing a pulse modulated waveform to better reflect electromyography (EMG) recordings of natural muscle activity [17]. Hart's algorithm introduces extra stimulation pulses in quick succession with a lowfrequency pulse train as shown in Fig. 7. These extra pulses or “doublets” greatly increase the strength of muscle contraction at specific points yet result in a much slower rate of muscle fatigue. C. Sensor Interface Block Most FES applications require some type of external control to determine when stimulus is required. Sensor inputs, in the form of force-sensitive-resistors (FSR), accelerometers, gyroscopes, tilt sensors, push buttons or EMG sensors, allow for control of the stimulation algorithm according to the wearer's physical state. For example, FSR sensors are used as foot switches to trigger stimulus of the peroneal nerve when the patient has lifted their heel, implementing a simple gait-phase detector [19]. The Bionic Glove senses voluntary wrist movement through the use of a displacement transducer and supplies electrical stimulation
Fig. 7. A stimulation envelope which mimics natural muscle activity by adding “doublet” pulse at specific intervals (Reprinted with permission).
to trigger hand opening and grasping; wrist flexion triggers hand opening, wrist extension triggers hand grasping [6]. Many FES devices, such as the FDA approved NESS H200, are open-loop systems that require the user's constant attention to operate through the use of push-button switches. Closed-loop FES devices use automatic control algorithms to reduce the necessity for user interaction for tasks such as balance while walking [20]. FSR foot switches are commonly used in dropped foot FES, devices such as the Odstock stimulators, the NESS L300 and the Duo-STIM make use of the in-shoe switches (usually fitted to an insole) to detect the heel strike or heel rise gait event. When a heel rise is detected, stimulation is applied to the peroneal nerve [19]. FSRs have a variable resistance, with applied pressure. The simplicity of the device (requiring little more than a voltage divider circuit), the small form factor and the low cost of foot switches are desirable qualities in a sensor, however, foot switches suffer from poor reliability. Due to the application of substantial, repeated forces to the sensor from its position under the heel of a patient, degradation of the resistor material and breakage of packaging often occurs which are obvious disadvantages for a user-friendly take-home device [3]. These issues have been minimised by careful packaging of the FSR device by Salisbury group [21]. However the identified disadvantages combined with the sensors' limited information [22] have encouraged researchers to investigate the use of other sensors for FES algorithm control, particularly MEMs based accelerometers and gyroscopes. Accelerometers are electronic devices capable of measuring the magnitude and detecting the direction of acceleration. Most accelerometers use a variation of the spring mass system [23], where when acceleration is applied, a small mass within the accelerometer responds by applying a force to a spring, causing it to stretch or compress. This displacement of the spring can be measured and used to calculate the applied acceleration [23]. A voltage is then output, proportional to the acceleration experienced. Due to advancements in integrated micro electro-mechanical systems (MEMS), low cost, low power accelerometers can be manufactured providing varying g ranges, such as ±2g, ±10g and ±50g etc. Most
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 18(2): 25-33, 2008© accelerometers today are tri-axial MEMS devices based on capacitative sensors (e.g. ADXL series, Analog Devices). Accelerometers have been suggested as a replacement for FSR foot switches (Fig. 8). Mansfield and Lyons placed an accelerometer on the lumbar region of 4 healthy subjects and determined that the accelerometer out performed foot switches in the number of correctly detected heel strike events [24]. Gyroscopes are used to measure angular velocity or rate
Fig. 8. (A) shows changes in AP horizontal acceleration during walking. The broken line shows the output from the footswitch (with 1 indicating ‘on’ and 0 indicating ‘off’) marking the beginning and end of each stride. Figure (B) shows the reduction of the acceleration curve presented in (A) into a pulse-like signal. A value of 1 indicates a negative–positive change (or increase) in the acceleration signal, while a value of 0 indicates a positive-negative change (or decrease) in the acceleration signal. The footswitch output is also shown for comparative purposes [24] (Reprinted with permission).
of rotation of the object to which they are attached. The rate of rotation is measured in reference to one of three axes, pitch yaw and roll. The most common method used by gyroscopes to measure angular velocity utilises the Coriolis effect [25]. MEMS gyroscopes contain a vibrating element in a reference frame. The use of gyroscopes to control FES algorithms has been investigated by several authors. Gyroscopes often make up part of a sensor array, being combined with accelerometers[26, 27] and FSRs [22]. Pappas et al. presented their gait-phase detection sensor (GPDS) in 2004. The GPDS contained a gyroscope, three FSR foot switches and a microcontroller embedded in a shoe insole. All signal processing of the GPDS took place in its microcontroller which outputted gait phases as discrete states to the Compex Motion FES device. This particular arrangement had a very high reliability, correctly applying stimulation 99% of the time on healthy and impaired-gait subjects. The work by Pappas et al. suggests that a combination of sensors may be the best approach to stimulus algorithm control. Many FES systems require the intervention of the patient or clinician to trigger stimulation or adjust stimulus levels. A more desirable system design could employ biofeedback to automatically adjust stimulus parameters or apply stimulation according to the wearer's physical state. EMG sensors have been integrated into various stimulator units to accomplish this task [28, 29]. Certain patients with tetraplegia retain the ability to extend and flex the wrist but
29
may have limited or no grasping function. The wrist movement can be captured using EMG by placing electrodes over the wrist extension muscle groups. Once a voluntary extension has passed a certain set EMG threshold, stimulation is triggered to activate the grasping function, thereby implementing EMG-triggered stimulation. Rather than just providing an on/off type stimulation trigger, the system used by Thorsen et al. was designed to apply a graduated stimulation intensity proportional to the amplitude of the captured EMG signal [28]. However, the authors noted that stimulation intensity was operating more like an on/off trigger, possibly due to a lack of fine tuning of the parameters. EMG sensors can be interfaced with the Compex Motion system. Popovic et al. used the Biofeedback Sensor model 2M4456 which consists of a 3 electrode surface EMG sensor for this purpose. The signal processing algorithms are embedded in the sensor’s microcontroller and are used to remove stimulation artefacts. The EMG sensors and signal processing algorithm can be used to trigger the stimulation output of the Compex Motion system [16]. The STIWELL med4 system (Otto Bock Healthcare, Duderstadt, Germany) can also be configured to use EMG sensors. Two of the device's stimulus channels double as EMG recording channels. The device makes novel use of the EMG signals by implementing such features as EMG-triggered stimulation, myosymmetry and EMG inhibition. Myosymmetry is used when a patient is unable to generate a measurable EMG signal on the affected limb, so the EMG sensors are placed over the unaffected limb instead and used to trigger stimulation in the affected limb. EMG inhibition is used to trigger stimulation only if the targeted muscles are contracted and aids the suppression of compensating muscle groups. The use of EMG as a sensing mechanism in an FES device requires algorithms to remove stimulus artefacts [30]. Tilt sensors measure the angle between a sensing axis and a reference vector such as gravity [31]. Various transforming elements can be used, such as a liquid inertial element or a pendulum or spring suspensions element. Tilt sensors have several advantages over foot switches as they can be built into the stimulator unit and do not suffer from degradation due to repetitive use. However, tilt sensors have a reduced response time compared to foot switches and are susceptible to acceleration artefacts adding a requirement for signal filtering. Rongching et al. investigated the suitability of four different types of tilt sensors in determining the phases of gait. One mercury based tilt sensor, two magnetoresistive tilt sensors and one electrolytic type tilt sensor were evaluated under the following criteria: physical size, mechanical reliability, signal stability and simplicity of conditioning the output signal. They eventually settled on using a Midori American Corp. (Fullerton, CA) UA-1 miniature magnetoresistive sensor due to its small form factor, low cost in comparison to the other sensors, simplicity of use (dc voltage divider set up) and reasonable signal quality (low-pass filtering can remove the acceleration component and the resonance component). This sensor was incorporated in a miniature, self-contained (consisting of no external wiring) foot drop FES device whereby stimulation was turned on or off based on the output signal of the tilt sensor rising or falling above or below predefined thresholds. The authors concluded that tilt
30
BRODERICK B.J., BREEN P.P., ÓLAIGHIN G.: ELECTRONIC STIMULATORS FOR SURFACE NEURAL PROSTHESIS photographs of suggested electrode placement sites.
Fig. 9. The WalkAide drop foot stimulator (Image courtesy of Innovative Neurotronics, Inc., TX).
sensors provide reliable stimulus control in response to gait events in real time. WalkAide drop foot stimulators (Fig. 9) are based on this control methodology [32]. D. User Interface Block and Programmability A user interface in the form of input buttons, LCD displays and PC-based GUIs is required to enable editing of selected stimulation parameters such as stimulation intensity and to provide feedback on system status. The user interface should be designed to suit the targeted end user where reduced mobility might make it difficult for a patient to adjust settings. The stimulator unit of the NESS H200 system features large push-button switches to trigger stimulation and a sliding control to adjust thumb flexion to accommodate different sized objects. The Parastep system makes use of buttons conveniently mounted on a walker to control ambulation and to assist the patient in changing from a standing position back to a sitting position. The DuoSTIM's control buttons were designed to be as ergonomic as possible for ease of use. The built-in LCD displays of the Duo-STIM, the UNAFET 8 and the Compex Motion systems display messages to the user such as stimulus intensity and battery charge level. The CEFAR REHAB X4 is the only FES device on the market that features a colour display. This increases the user friendliness of the device by allowing access to program guidelines and displaying
Fig. 10. Various screenshots of the Duo-STIM programmer interface (Reprinted with Permission).
More and more FES devices are being designed to allow a programmable interface between the stimulator device and a PC graphical user interface (GUI) or some other custom programming unit. For example, a GUI designed for Windows (WalkAnalyst) allows the clinician to set and adjust stimulus strength and timing parameters on the WalkAide device. This separate programming system is deigned in order to restrict access to certain modifiable stimulation parameters exclusively for clinician use and to reduce the perceived complexity of the device for the patient. The user is often only required to access a minimum control set, such as adjusting stimulation intensity. Stimulation algorithms, waveform shapes, pulse durations and sensor thresholds are commonly adjusted via the programming unit. Chosen parameters and algorithms can be downloaded onto the device or stored for later use. This offers a huge benefit over hard-wired devices where the interface to adjust these advanced stimulation settings must be built into the stimulator unit itself with implications for the required complexity of the device controls and clinicians are usually limited to adjusting stimulation intensity or frequency.
Fig. 11. A screenshot of the main screen of the Compex Motion GUI highlighting features such as (1) setup functions, (2) stimulation mode functions, (3) stimulation frequency setting, (4) memory chip-card functions, (5) time line editing functions for stimulation channel No. 3, (6) pulse amplitude for stimulation channel No. 2, (7) upper limits for pulse width and amplitude for stimulation channel No. 1, (8) time line for stimulation channel No. 2, and (9) primitive “jump to first” in time line No. 1. [16] (Reprinted with Permission).
The fully programmable nature of the Duo-STIM (Fig. 10) and the Compex Motion (Fig. 11) systems allows these devices to be potentially adapted to a wide range of FES applications without modifications to the hardware. The Duo-STIM's programmer unit is a battery powered device incorporating a touch screen interface which allows a clinician to select the algorithm to be executed by the stimulator and adjust the algorithm's parameters such as maximum stimulus intensity, stimulation pulse width and frequency. These parameters are downloaded to the stimulator unit via a RS-232 interface. The flexibility of the Compex Motion stimulator comes from its PC based GUI. The GUI is designed to facilitate the programming of each of the 4 stimulation channels separately. Drag and drop
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 18(2): 25-33, 2008© functionality is employed whereby a user can select 56 icons or primitives, consisting of ramps, loops, pulse widths etc., each one describing a different task the stimulator can perform. This system allows the user to take full advantage of the flexibility of the system. Completed programs are downloaded onto programmable data cards which are then inserted into the stimulator, instantly changing the stimulators function. The GUI allows FES practitioners to store user parameters as binary files on their computers where they can be later edited or reused[14, 16]. E. Power Supply Block FES devices designed for take-home should be designed to maximize battery life and minimize battery recharging times to reduce the amount of time the patient must be without the device. Due to the large output voltage required by the Parastep FES system, a separate, belt-worn, battery power pack made up of nickel cadmium cells was necessary to power the system, which only held their charge for 1 hr. 45 min. and took several hours to recharge. Newer rechargeable FES systems have used nickel-metal hydride batteries, allowing up to 8 hours of continuous stimulation [14], and lithium ion batteries, allowing over 10.5 hours of stimulation [15], both requiring a 2 hour recharge time. Some designs, such as the WalkAide, use disposable batteries which reduces the safety considerations of the device by removing the recharge circuitry. F. Encapsulation Portability is concerned with the physical size and weight of the device, its ergonomics, its physical appearance and its ease of donning. As many people are familiar with mobile phones and portable music players such as the iPod (which typically have had tens to hundreds of millions of dollars invested in the device design), the expectation is that commercial FES devices should be as aesthetically pleasing and be of a similar size to these mass market devices if they are to be used on a day to day basis (while the typical budgets available for the design of commercial FES devices is a small fraction of those other budgets). A device that is aesthetically pleasing is more likely to have better patient compliance, especially if its use is required in a public setting [33]. It is of utmost importance that the patient feels
Fig. 12. NESS L300 drop foot stimulator and wireless foot switch (Image courtesy of Bioness Inc., CA).
31
emotionally positive about using the FES device. There has been a move away from “box on the hip” style stimulators to a more “wearable electronics” style, with such devices as the WalkAide and the NESS L300 (Fig. 12). These devices are placed directly over the stimulation site (the common peroneal nerve) and contain no electrode leads or sensor wires. Although the NESS H200 does have a separate hip worn control unit, it combines FES with an orthosis to improve hand function, following the style of wearable electronics (Fig. 13). The orthosis is worn over the forearm and holds the wrist at a fixed functional angle. Five electrodes, individually positioned for each patient, are attached to the inside shell of the orthosis. The electrode pads are humidified with tap water for better conductivity. This allows for very quick and easy donning of the equipment, requiring little to no assistance.
Fig. 13. NESS H200 is a hybrid orthosis-FES device. The orthosis is worn over the arm and hand. Stimulus is supplied by the control unit (Image courtesy of Bioness Inc., CA).
G. Safety Safety is a major concern with FES systems as its primary function is to apply electrical energy directly to a patient. To ensure patient safety, devices must be designed so that if any component in the device fails it fails in a safe manner. Special safety checks implemented in hardware, software or both should prevent any injury to the patient. This is so called fail-safe design. The stimulator unit of the system should employ some interlocking system circuitry which disables the stimulation output stage while the unit is connected to the mains supply for battery charging. This will prevent a situation arising where a direct connection is made between the mains supply and the patient. Certain safety features were incorporated into the Parastep device such as an alert system to warn if an electrode is accidentally disconnected from the skin, an implementation of the fail-safe philosophy. The system then handles the error by preventing all functions of the microcomputer from executing until the electrodes are corrected. If the patient is standing and a disconnected electrode is detected, all functions will also cease, except
32
BRODERICK B.J., BREEN P.P., ÓLAIGHIN G.: ELECTRONIC STIMULATORS FOR SURFACE NEURAL PROSTHESIS
stimulation to the quads to allow the patient to return to the sitting position to correct the error. To ensure user safety, the Duo-STIM stimulator unit will not operate while it is being charged through the mains supply. The rechargeable battery itself incorporates a safety IC to prevent damage to the battery if shorted which could, under these circumstances, explode causing injury. The firmware of the Duo-STIM also executes a serious of safety checks on the hardware of the device during its setup procedures. If any of these checks fail, the firmware outputs a warning message on the LCD screen and will not allow an algorithm to execute until the problem is resolved.
III. DESCUSSION It is not always appropriate to implant a device immediately after injury. Instead, surface FES devices should be first used on patients before deciding upon an implanted system [6, 9]. This would allow patients time to get used to having a neuroprosthesis and help them learn the control features of the device which may improve acceptance of implanted FES. It would also provide some initial motor relearning, the benefits of which would be encouraging for the patient. Hence, it may be even more beneficial for the surface FES device to be compatible with the implanted system, by having for example, a similar user interface, to allow easier migration over to the implanted system. The earlier the surface stimulator is introduced in the rehabilitation after injury, the greater the likelihood of a successful return of motor function and the faster the recovery will be [9]. The benefits to patients from using FES devices are well known. Several comprehensive studies have evaluated the use of surface FES devices and reported significant improvements in function such as upper limb grasping, standing/stepping and ankle dorsiflexion. The promising standing and walking performance [4, 34], positive cardiopulmonary responses [4] and increase in lower extremity blood flow [35] caused by the Parastep have proved that it is significantly beneficial to SCI patients. Studies performed by Prochazha et al., Popovic et al. and Alon et al. showed improves in upper limb function such as tenodesis grasp and grasping power [6, 36] and improvements in the number of activities of daily living performed [37]. Taylor et al. found the Odstock dropped foot stimulator improved walking speed and reduced physiological cost index in stroke patients suffering from drop foot [19]. Although FES devices are usually aimed at SCI patients or patients suffering from dropped foot, their applications are not limited to these patient groups. Nathan et al. and Stein et al. mention that patient groups, such as cerebral vascular accident, head injury, cerebral palsy, multiple sclerosis, hemiparesis following surgery, familial spastic paraparesis and upper limb orthopaedic patients, have also made use of FES devices [32, 38]. Due to the time consuming and costly nature of validating and testing a medical device, few FES devices have made it past the research stage in university labs to reach a wider market. Although numerous devices have been developed, the majority were only designed to carry out a specific task and were only applied to a small number of people [16].
Of course FES systems can only be effective in facilitating functional and therapeutic stimulation if patients opt to use them. Patients may be reluctant to try or continue using FES for long periods if there is difficulty donning and doffing the equipment. This presents a unique design requirement as effective devices, especially for upper limb applications, should be capable of being donned and doffed by the patient with little or no assistance despite the patient having reduced dexterity. Overly complicated user interfaces and large, bulky designs can deter patients from using the device on a day to day basis. Wired sensors would also certainly hinder patient acceptance especially in every day situations as they can be uncomfortable and unsightly. Therefore, we believe FES devices should incorporate wireless or built-in sensors and be mounted directly over stimulation sites to remove the necessity for electrode leads in an effort to increase patient acceptance of FES systems.
REFERENCES [1] [2]
[3]
[4]
[5]
[6]
[7] [8]
[9]
[10]
[11]
[12] [13]
[14]
[15]
[16]
D. N. Rushton, "Functional Electrical Stimulation," Physiological Measurement, vol. 18, pp. 241-275, 1997. W. T. Liberson, H. J. Homquest, D. Scott, and M. Dow, "Functional electrotherapy in stimulation of the peroneal nerve synchronized with the swing phase of gait in hemiparetic patients," Archices of Physical Medicine and Rehabilitation, vol. 42, pp. 202-205, 1961. G. M. Lyons, T. Sinkjær, J. H. Burridge, and D. J. Wilcox, "A Review of Portable FES-Based Neural Orthoses for the Correction of Drop Foot," IEEE Trans Neural Syst Rehabil Eng, vol. 10, pp. 260-79, 2002. D. Graupe and K. H. Kohn, "Functional neuromuscular stimulator for short-distance ambulation by certain thoracic-level spinal-cordinjured paraplegics," Surgical Neurology, vol. 50, pp. 202-207, 1998. P. H. Peckham and J. S. Knutson, "Functional Electrical Stimulation for Neuromuscular Applications," Annual Review of Biomedical Engineering, vol. 7, pp. 327-360, 2005. A. Prochazka, M. Gauthier, M. Wieler, and Z. Kenwell, "The bionic glove: An electrical stimulator garment that provides controlled grasp and hand opening in quadriplegia," Archives of Physical Medicine and Rehabilitation, vol. 78, pp. 608-614, 1997. L. R. Sheffler and J. Chae, "Neuromuscular electrical stimulation in neurorehabilitation," Muscle & Nerve, vol. 35, pp. 562-590, 2007. G. M. Lyons, G. E. Leane, M. Clarke-Moloney, J. V. O'Brien, and P. A. Grace, "An investigation of the effect of electrode size and electrode location on comfort during stimulation of the gastrocnemius muscle," Medical Engineering & Physics, vol. 26, pp. 873-878, 2004. M. R. Popovic, T. Keller, I. P. I. Papas, V. Dietz, and M. Morari, "Surface-stimulation technology for grasping and walking neuroprostheses," Engineering in Medicine and Biology Magazine, IEEE, vol. 20, pp. 82-93, 2001. M. Ilic, D. Vasiljevic, and D. B. Popovic, "A programmable electronic stimulator for FES systems," Rehabilitation Engineering, IEEE Transactions on, vol. 2, pp. 234-239, 1994. L. L. Baker, D. R. McNeal, L. A. Benton, B. R. Bowman, and R. L. Waters, Neuromuscular Electrical Stimulation - A Practical Guide, 3rd ed. Downey California: Rancho Los Amigos Research and Education Institute, 1993. J. P. Reilly, Electrical Stimulation and Electropathology. Cambridge: Cambridge University Press, 1992. G. Kantor, G. Alon, and H. Ho, "The effects of selected stimulus waveforms on pulse and phase characteristics at sensory and motor thresholds," PHYS THER, vol. 74, pp. 951-962, October 1, 1994 1994. T. Keller, M. R. Popovic, I. P. I. Pappas, and P. Y. Müller, "Transcutaneous Functional Electrical Stimulator "Compex Motion"," Artificial Organs, vol. 26, pp. 219-223, 2002. P. P. Breen, G. J. Corley, D. T. O'Keeffe, R. Conway, and G. ÓLaighin, "A programmable and portable NMES device for drop foot correction and blood flow assist applications," Medical Engineering & Physics, vol. In Press, Corrected Proof, 2008. M. R. Popovic and T. Keller, "Modular transcutaneous functional electrical stimulation system," Medical Engineering & Physics, vol. 27, pp. 81-92, 2005.
JOURNAL OF AUTOMATIC CONTROL, UNIVERSITY OF BELGRADE, VOL. 18(2): 25-33, 2008© [17] D. J. Hart, P. N. Taylor, P. H. Chappell, and D. E. Wood, "A microcontroller system for investigating the catch effect: Functional electrical stimulation of the common peroneal nerve," Medical Engineering & Physics, vol. 28, pp. 438-448, 2006. [18] P. P. Breen, D. T. O'Keeffe, R. Conway, and G. M. Lyons, "A system for the delivery of programmable, adaptive stimulation intensity envelopes for drop foot correction applications," Medical Engineering & Physics, vol. 28, pp. 177-186, 2006. [19] P. N. Taylor, J. H. Burridge, A. L. Dunkerley, D. E. Wood, J. A. Norton, C. Singleton, and I. D. Swain, "Clinical use of the odstock dropped foot stimulator: Its effect on the speed and effort of walking," Archives of Physical Medicine and Rehabilitation, vol. 80, pp. 15771583, 1999. [20] C. L. Lynch and M. R. Popovic, "Functional Electrical Stimulation," Control Systems Magazine, IEEE, vol. 28, pp. 40-50, 2008. [21] I. D. Swain, J. H. Burridge, S. E. Crook, S. M. Finn, C. A. Johnson, G. E. Mann, P. N. Taylor, and P. A. Wright, "From circuit design to service delivery—establishing a clinical FES service," in Proc. 6th Ann. IFESS Conf., 2001, pp. 117–119. [22] I. P. I. Pappas, T. Keller, S. Mangold, M. R. Popovic, V. Dietz, and M. Morari, "A reliable gyroscope-based gait-phase detection sensor embedded in a shoe insole," Sensors Journal, IEEE, vol. 4, pp. 268274, 2004. [23] M. J. Mathie, A. C. F. Coster, N. H. Lovell, and B. G. Celler, "Accelerometry: providing an integrated, practical method for longterm, ambulatory monitoring of human movement," Physiological Measurement, vol. 25, pp. R1-R20, 2004. [24] A. Mansfield and G. M. Lyons, "The use of accelerometry to detect heel contact events for use as a sensor in FES assisted walking," Medical Engineering & Physics, vol. 25, pp. 879-885, 2003. [25] R. P. Leland, "Adaptive mode tuning for vibrational gyroscopes," Control Systems Technology, IEEE Transactions on, vol. 11, pp. 242247, 2003. [26] S. Simcox, S. Parker, G. M. Davis, R. W. Smith, and J. W. Middleton, "Performance of orientation sensors for use with a functional electrical stimulation mobility system," Journal of Biomechanics, vol. 38, pp. 1185-1190, 2005. [27] K. Tong, A. Mak, and W. Ip, "Command control for functional electrical stimulation hand grasp systems using miniature accelerometers and gyroscopes," Medical and Biological Engineering and Computing, vol. 41, pp. 710-717, 2003. [28] R. Thorsen, M. Ferrarin, R. Spadone, and C. Frigo, "Functional Control of the Hand in Tetraplegics Based on Residual Synergistic EMG Activity," Artificial Organs, vol. 23, pp. 470-473, 1999. [29] S. Saxena, S. Nikolic, and D. Popovic, "An EMG-controlled grasping system for tetraplegics," Journal of Rehabilitation Research & Development, vol. 32, p. 17, 1995. [30] D. T. O'Keeffe, G. M. Lyons, A. E. Donnelly, and C. A. Byrne, "Stimulus artifact removal using a software-based two-stage peak detection algorithm," Journal of Neuroscience Methods, vol. 109, pp. 137-145, 2001. [31] D. Rongching, R. B. Stein, B. J. Andrews, K. B. James, and M. Wieler, "Application of tilt sensors in functional electrical stimulation," Rehabilitation Engineering, IEEE Transactions on, vol. 4, pp. 63-72, 1996. [32] R. B. Stein, S. Chong, D. G. Everaert, R. Rolf, A. K. Thompson, M. Whittaker, J. Robertson, J. Fung, R. Preuss, K. Momose, and K. Ihashi, "A Multicenter Trial of a Footdrop Stimulator Controlled by a Tilt Sensor," Neurorehabil Neural Repair, vol. 20, pp. 371-379, 2006. [33] M. Diaz, "Emotions in Assistive Technologies Design for Elders," in First International Congress on Digital Homes, Robotics and Telecare for All Barcelona, Spain, 2005. [34] K. J. Klose, P. L. Jacobs, J. G. Broton, R. S. Guest, B. M. NeedhamShropshire, N. Lebwohl, M. S. Nash, and B. A. Green, "Evaluation of a training program for persons with SCI paraplegia using the Parastep®1 ambulation system: Part 1. Ambulation performance and anthropometric measures," Archives of Physical Medicine and Rehabilitation, vol. 78, pp. 789-793, 1997. [35] M. S. Nash, P. L. Jacobs, B. M. Montalvo, K. J. Klose, R. S. Guest, and B. M. Needham-Shropshire, "Evaluation of a training program for persons with SCI paraplegia using the Parastep®1 ambulation system: Part 5. Lower extremity blood flow and hyperemic responses to occlusion are augmented by ambulation training," Archives of Physical Medicine and Rehabilitation, vol. 78, pp. 808-814, 1997. [36] D. Popovic, A. Stojanovic, A. Pjanovic, S. Radosavljevic, M. Popovic, S. Jovic, and D. Vulovic, "Clinical evaluation of the bionic glove," Archives of Physical Medicine and Rehabilitation, vol. 80, pp. 299-304, 1999.
33
[37] G. Alon, K. McBride, and H. Ring, "Improving selected hand functions using a noninvasive neuroprosthesis in persons with chronic stroke," Journal of Stroke and Cerebrovascular Diseases, vol. 11, pp. 99-106, 2002. [38] R. H. Nathan, H. P. Weingarden, A. Dar, and C. Macaspac, "Restoration of Functional Activities with the Handmaster NMS1," in Proc 6th Int Workshop Functional Electrostimulation Vienna, Austria, 1998, pp. 251-254.
