A hybrid neuroprosthesis (HNP)1–10 combines mechanical orthoses and functional neuromuscular stimulation (FNS) to restore ambulatory function to individuals paralyzed by spinal cord injury (SCI). For walking after paralysis, the purpose of traditional bracing11–14 is to keep the user upright and prevent collapse. Accordingly, the knees and ankle joints are constrained in extension and neutral, respectively. With an isocentric reciprocal gait orthosis (IRGO), the hips are reciprocally coupled so that extension of one hip causes flexion of the other, thus maintaining upright posture when both feet are on the ground.14 Sagittal limb movements that bring about forward progression for walking with an IRGO are induced by the coordinated actions of the upper limbs on a walking aid.15 Thus, bracing is associated with high user effort during walking,16 which limits its functional utility.17 Alternatively, FNS systems18–24 deliver electrical pulses to the nerves below the level of injury to elicit contractions in the lower limb muscles for support against collapse and power for forward progression of the user. Because the electrically stimulated paralyzed muscles fatigue quickly,25 their ability to support the body against collapse diminishes and compromises walking distance.21 In addition, significant upper limb effort is required to maintain trunk posture for FNS-only gait.16
The goal of HNP systems is to combine the advantages and minimize the shortcomings of either bracing or FNS by themselves to create a more effective intervention. Lower limb bracing reduces the degrees of freedom, minimizes the secondary actions of muscles stimulated by FNS, and simplifies control. Functional neuromuscular stimulation provides propulsive power during locomotion and eliminates the high energy compensatory mechanisms commonly used for brace walking.16
The HNP approach divides walking into two main functional tasks: body weight support (accomplished by bracing) and limb and body propulsion (accomplished by FNS). First, the supportive tasks provided by bracing include weight bearing, maintaining trunk posture, and guiding limb trajectories. Because stimulation can be turned off when support is provided with bracing, the reduction of stimulation duty cycle should potentially delay the onset of fatigue of the stimulated paralyzed muscles.26 Furthermore, maintaining upright posture with bracing should lessen upper limb loading and delay the onset of fatigue of the voluntary muscles. Second, when selected constraints of the brace are removed (i.e., unlocking the knee joint), FNS of the target muscles can advance the leg during swing without upper limb compensatory mechanisms as required during stiff-legged brace walking. The successful outcome of both tasks should result in reduced upper body loading on the walking aid and improved mobility relative to walking with bracing alone and reduced muscle stimulation and upper limb support compared with walking with FNS-only systems.
In this study, we designed a sensor-based controller to modulate stimulation to knee extensor muscles synchronized with the state of a stance control knee mechanism (SCKM).27 The controller minimizes stimulation to extend the knee during FNS-induced swing in preparation for heel strike and during stance when knee support against collapse is provided by the SCKM. It was hypothesized that the HNP would significantly reduce stimulation duty cycle as compared with FNS-only walking through sensor-based control of stimulation to the knee extensors and stabilization of the knee during stance by means of the SCKM. In addition, it was hypothesized that upper body forces could be significantly reduced as compared with FNS-only walking by maintaining upright posture with bracing. Furthermore, the HNP would reduce limb effort by obviating the need for compensatory mechanisms to advance the leg during swing as compared with brace-only walking. To assess the benefits of this HNP, the upper limb forces applied to the walking aid and the duty cycle of knee extensor muscle stimulation of an individual with paraplegia walking with the HNP were measured and compared with walking with either IRGO or FNS alone. The effect of the SCKM itself including mass and inherent viscous and frictional properties on the knee trajectory during swing was also examined.
A 70-kg man with T-9, ASIA A class SCI consented to participate in this study as approved by the local institutional review board. He had US Food and Drug Administration–approved (investigational device exemption: GS900107) percutaneous intramuscular electrodes implanted near nerves innervating most major muscles controlling the hips, knees, and ankles, including iliopsoas for hip flexion,28 tensor fasciae latae, sartorius and gracilis for hip and knee flexion, posterior portion of adductor magnus, hamstrings and gluteus maximus for hip extension, quadriceps for knee extension, tibialis anterior and peroneus longus for ankle dorsiflexion, and gastrocnemius for ankle plantarflexion.21 Baseline stimulus patterns for FNS-only walking and standing were generated heuristically following previously established tuning rules.29 The participant has had more than 24 years of experience with his FNS system for exercise and standby assisted walking.30 He was also proficient in walking with an IRGO with or without stimulation.
The HNP evaluated in this study included a modified IRGO with SCKM27 and an FNS subsystem consisting of percutaneously implanted electrodes and microprocessor controlled external stimulator. An IRGO14 that reciprocally coupled hip flexion with contralateral hip extension in a 1:1 ratio was used to provide postural support. The IRGO was modified to include adjustable lateral uprights with straps below the knees, across pelvis, and at chest level with standard abduction hip joints for easy donning while seated in the wheelchair.30 Ankles were mechanically locked at neutral.
Stance control knee mechanisms were incorporated into the modified IRGO to replace standard drop lock knee joints. The function of the SCKM was to lock the knee during stance and unlock it to provide foot-ground clearance during swing (Figure 1). The SCKM is a hydraulic mechanism incorporating a single-rod, double-acting cylinder (Clippard Minimatic, Cincinnati, OH, USA) with a 9/16-in bore, 0.25-in rod diameter, and 3-in stroke rated at the maximum operating pressure of 2000 psi. Inline between the cylinder’s ports is a two-way, two-position normally closed solenoid valve (Allenair Corp, Mineola, NY, USA) that allows the knee to lock without power during the standing and stance phase of gait. A miniature single-acting cylinder with spring return was incorporated into the system to act as an accumulator to take up the volume differential between the blind and rod sides of the locking cylinder. As a safety measure, the mechanism was designed to lock only against knee flexion to prevent collapse and always allow extension to maintain upright stance. This was achieved by specifying the valve cracking pressure to be low enough to be easily exceeded by applying a small extension moment with FNS. The cylinder was attached by rod eye spherical and clevis joints to the thigh and leg uprights, respectively, to convert linear movement of the cylinder’s piston to rotary movement of the knee. When fully extended in bench testing, the SCKM resisted up to 70 Nm of torque, well above what is normally observed during gait. Thus, the mechanism would be capable of providing support when stimulation is turned off during stance. Equally important, the SCKM reliably unlocked within 200 milliseconds under knee flexion torque of up to 50 Nm and in 12 milliseconds when unloaded. The passive resistance was less than 2 Nm. Compliance of the mechanism, defined as the change in angle when the SCKM is locked, increased with applied flexion torque from about 2° to 5° of knee flexion.27
The FNS subsystem provided the power for moving the lower limbs and the body forward during walking. Constant current charge balanced biphasic pulses at 20 mA and stimulus pulse width up to 250 microseconds were delivered to keep within commonly accepted safe charge limits.31 Maximum stimulus pulse widths were determined for each electrode by manual testing to achieve the highest strength with minimal recruitment of adjacent muscles as determined by palpation.29 The minimum stimulus interpulse interval was limited to 30 milliseconds for a maximum frequency of 33.3 Hz to minimize muscle fatigue.32 A sensor-based controller was designed to modulate the stimulation of knee extensor muscles, whereas baseline preprogrammed stimulation was used for control of the hip and ankle.
SENSOR-BASED FUNCTIONAL NEUROMUSCULAR STIMULATION CONTROLLER
The FNS controller consisted of two modules: 1) a gait phase module, which generated output signals indicating the timing of phases of the gait cycle used to synchronize constraints of the brace with FNS activity, and 2) the knee FNS control module, responsible for the real-time modulation of the stimulation based on knee angle and SCKM constraint state. All software for the FNS controller was developed and implemented in the Matlab®/Simulink/xPC Target programming environment (The MathWorks, Inc, Natick, MA, USA).
GAIT PHASE MODULE
This module sent out timing signals derived from the baseline stimulation pattern of hip and knee flexors and knee extensors, which indicated the occurrence of specific phases in the gait cycle for each leg. Two gait phases were designated specifically for the proper and timely operation of the SCKM. Phase 1 represents the double-support preswing phase of gait, during which a burst of knee extensor stimulation effectively unloaded the SCKM for quicker unlocking. Phase 2 represents the swing phase of gait and was implemented to prevent the SCKM from locking during swing. Without these phase timing signals derived from the baseline stimulation pattern, the SCKM would lock upon foot contact with the ground, based on feedback from force-sensitive resistors (B & L Engineering, Tustin, CA, USA) placed under the soles of the feet.33 The SCKM controller ensured that the knee joint would remain unlocked as long as the limb was in phase 2, thus preventing locking of the SCKM during swing from faulty signal or from toe dragging from insufficient foot-to-ground clearance.
The phase timing output signals were derived from the baseline stimulation activity of hip and knee flexors and knee extensors (Figure 2). The period of phase 1 was set to begin 210 milliseconds before the onset of hip (iliopsoas, sartorius) and knee flexor (sartorius) stimulation and to end at the onset of hip and knee flexor stimulation to initiate swing. The period of phase 2 was set to begin 120 milliseconds before the onset of hip and knee flexor stimulation and to end 210 milliseconds after the onset of knee extensor stimulation in preparation for heel strike. The onset time of phase 2 was timed to allow the SCKM to unlock before hip and knee flexor activation for swing. The timing of the two phases prevented an excessive flexion moment on the SCKM, which would prolong unlocking.27 This accounted for the unlocking response time of the SCKM and prevented the stimulated hip and knee flexors from working against a locked SCKM, which would impair the swing trajectory and prevent proper foot-to-ground clearance. The end of phase 2 was after knee extensor stimulation (quadriceps) onset to allow the swing limb to fully extend in preparation for heel strike before SCKM locking occurred.
KNEE FUNCTIONAL NEUROMUSCULAR STIMULATION CONTROL MODULE
The knee FNS control module coordinated the activation/deactivation of knee extensors with the unlocking/locking of the SCKM. As a result, all stimulation to knee extensors was turned off when the knee was extended and the user’s body weight supported by the SCKM, significantly reducing the muscle stimulation duty cycle.26,34 Stimulus pulse width and interpulse interval to the knee extensors were modulated relative to baseline levels by the following two rules.
Rule 1: If the knee is fully extended, locked against flexion, and not in preswing phase 1, then deactivate stimulation to the stance knee extensors. The knee was considered to be fully extended when the knee angle was between two specified values (threshold 1 = 3° and threshold 2 = 13°). A precision rotary potentiometer (Vishay, Malvern, PA, USA) on each SCKM was used to measure knee angle.
Threshold 1 was selected to prevent reactivation of preset stimulation levels if the knee angle changed because of the mechanical compliance in the SCKM.33 Threshold 2 was specified as the acceptable knee flexion angle when the SCKM was locked. The knee being locked against flexion was determined from the signal that controlled the state of the SCKM valve. When the valve was closed, the knee was locked against flexion. The SCKM being both extended and locked against flexion indicated that the limb was in stance and could fully support the user against collapse. A burst of stimulation was applied to the knee extensors to reduce the flexion torque applied to the SCKM before unlocking the mechanism during preswing phase 1. This was done to ensure that the SCKM would unlock responsively with minimal delay.
Rule 2: If the knee transitions to a locked state but has not reached full extension, then increase knee extensor stimulation frequency to 33.3 Hz until the knee has fully extended. Locking of SCKM indicated that the foot had contacted or was about to contact the ground.33 Once locked, the SCKM resisted knee flexion; however, it allowed for free knee extension at all times so the knee can be driven into full extension with an increased stimulus frequency until threshold 1 is reached. If the compliance of the mechanism caused stance phase knee flexion to exceed threshold 2, then stimulation was reactivated to drive the knee back into extension past threshold 1, at which time the stimulation would be turned off. Because of software restrictions, interpulse interval modulation was limited to switching between 60 and 30 milliseconds, corresponding to a stimulus frequency of 16.7 and 33.3 Hz, respectively. The stimulus pulse width was not increased along with frequency to prevent over stimulating the quadriceps muscles.
The FNS controller was prototyped and implemented using 1) a target personal computer (PC), 2) a host computer, and 3) the computer controlled stimulator. The target PC ran the xPC target kernel that implemented the HNP controller and acquired sensor data in real time. The HNP controller consisted of the SCKM controller33 and FNS controller and performed zero calibration and low-pass filtering of the sensor signals. The target PC was equipped with data acquisition boards (National Instruments, Austin, TX, USA) with multiple analog and digital I/O channels for sampling sensor signal and outputting control signals to the SCKM. Twenty meters of shielded multiconductor cabling connected the instrumented modified IRGO to the target PC. All communication between the target PC and HNP was at a frequency of 200 Hz. The host PC communicated with the target PC through a custom Matlab graphical user interface for the building and implementation of the HNP controller on the target PC. Up to 24 independent channels of constant current biphasic charge-balance asymmetric stimulus pulses at the specified time, pulse width, and interpulse interval were generated and delivered by the stimulator.35 A 20-m, 550-MHz patch cable connected the stimulator to the target PC, which updated the instantaneous stimulus parameters (i.e., pulse width, interpulse interval, and current amplitude) every 30 milliseconds. The stimulator was powered by an internal Sony 15.8 Wh NP-F570 7.2–8.4 V lithium ion rechargeable battery pack (Sony Corporation, Tokyo, Japan).
HYBRID NEUROPROSTHESIS EVALUATION
A 16-camera Vicon MX40 (Vicon, Inc, Oxford, UK) motion analysis system (sampling at 100 Hz) surrounded an 8 × 3 × 2–m work volume to record the three-dimensional coordinates of reflective markers placed at key locations on the orthosis and bony landmarks of the body. Nineteen markers were placed on the participant,36 with 15 of these marker placements modified to accommodate the presence of the bracing.37 The marker coordinates were used to calculate the trunk and joint kinematics, speed, cadence, and step length. A digital pulse was sent from the target PC to the Vicon workstation to synchronize the data collected by the separate data acquisition systems.
Load cells (AMTI, Inc, Watertown, MA, USA) were attached to each handle of a walker to measure the vertical components of the forces applied to the assistive device by the arms. The load cell signals were low-pass filtered online (seventh-order Butterworth) at a cutoff frequency of 20 Hz.
The upper limb loads applied to the walker during gait with the HNP were compared with three control cases: walking with 1) FNS only; 2) IRGO only, with hips reciprocally coupled, knees locked in extension, and ankles locked at neutral; and 3) IRGO with hips reciprocally coupled, knees freed (i.e., drop locks of the standard knee-ankle-foot orthosis unlatched), ankles locked at neutral, and joints driven by preprogrammed muscle stimulation pattern (i.e., HNP0). For FNS-only walking, a baseline stimulation pattern was used to control hip flexion and extension, knee flexion and extension, and ankle plantar and dorsiflexion. For HNP0, a scaled down baseline stimulation pattern was used for activation of hip flexors and extensors and knee flexors and extensors. This baseline stimulation pattern was the same for HNP, except that stimulation of knee extensors was modulated with sensor-based controller synchronized with SCKM constraints. The influence of the FNS controller on stimulation duty cycle of the knee extensors was evaluated relative to the established baseline stimulation patterns. The effect of the SCKM in the HNP on knee motion was evaluated against that of the HNP0 system.
For each test case, six trials were performed, with the participant instructed to walk through the work volume at his preferred speed. For FNS-only walking, stimulation for stepping was cycled continuously through the baseline pattern at his preferred speed after being initiated by a handheld switch. For the HNP and HNP0, each step was triggered manually via the finger switch at the preferred pace set by the user. All data from repeated strides under each condition were normalized to percentage gait cycle and averaged. The gait cycles were determined from the minimum vertical coordinate of the heel markers separately for each leg. The mean and average maximum upper limb loading were determined over approximately 25 strides for each test case. Analysis of variance with 95% confidence interval (p < 0.05) was used to determine statistically significant differences between walking systems. A Student t test was used when comparing variables between two walking systems.
Figure 3 shows snapshots of the participant walking with the HNP system. The average forward lean of the trunk for the HNP was 3° and 19° for FNS-only walking (p < 0.001). The sagittal movement of the trunk during gait with the HNP fluctuated by 21% (p < 0.001) less than when walking with the IRGO only.
The average gait speed with the HNP (Table 1) was not statistically different from walking with the HNP0 (p = 0.913) and was significantly faster than walking with the IRGO only (p = 0.017). However, the average gait speed of the HNP was significantly slower than that observed in the FNS-only case (p < 0.001), with shorter mean step lengths (left, p < 0.001; right, p = 0.131). The hip flexion with HNP was less than 40° as compared with 74° in the FNS-only walking. Similarly, the step lengths with the HNP0 were shorter than in FNS-only walking (left, p < 0.001; right, p = 0.118). With FNS-only walking, step length was more symmetric than when walking with the orthosis alone or combined with FNS where the right step was generally longer than the left.
The duration of each step was predefined by the preprogrammed stimulation pattern that cycled at a fixed rate for FNS-only walking, whereas manually triggered stepping with FNS combined with conventional IRGO was more variable as the user introduced a delay between successive steps. Thus, the cadence was significantly lower than that for the FNS-only walking (left, p < 0.001; right, p < 0.001) and similar to walking with IRGO only. The left step cadence for all cases involving bracing was lower and more variable than the right.
UPPER LIMB LOADING
Figure 4 shows the mean and average maximum of the total vertical upper limb force on the walker for each walking system normalized by the body weight of the subject. The orthotic component of the HNP is essentially self-supportive, where the weight is transmitted to the ground by means of lateral uprights. To show the absolute contribution of the HNP to the overall reduction of the upper limb effort, the weight of the orthosis was not added to the participant’s body weight when normalizing the upper limb forces. The mean upper limb force while walking with the HNP was 36% less than when walking with FNS only (p < 0.001), although average maximum forces were similar (p = 0.50). In addition, the mean and average maximum forces for walking with the HNP were 17% (p < 0.001) and 30% (p < 0.001), respectively, less than for walking with the IRGO only. However, the mean and average maximum upper limb forces for the HNP were 8% (p = 0.012) and 16% (p = 0.001), respectively, higher than those for the HNP0 walking. These results show that with HNP, the upper limb load on the walker was reduced as compared with IRGO-only and FNS-only gait regardless of the configuration of the orthotic components and controlled joint locking.
MUSCLE DUTY CYCLE
Figure 5 shows the knee angle and preprogrammed stimulation pattern pulse widths applied to the quadriceps for FNS-only and HNP0 walking and stimulation pulse width modulated by the FNS controller when walking with the SCKM-based HNP. The muscle duty cycle was calculated by dividing the stride stimulation time by the duration of the stride. The average stride duration was 4.08 ± 1.05 seconds for walking with the HNP. With baseline stimulation, the average stimulation ON time (time the muscle is being activated by the FNS) for both the left and right quadriceps was approximately 3.41 ± 1.04 seconds per stride. With the sensor-based FNS controller, stimulation to the quadriceps was applied for 1.06 ± 0.41 seconds, with 3.02 ± 0.82 seconds of rest per stride. Thus, the duty cycle with HNP was reduced to 26% ± 6% from 82% ± 5% when walking with FNS only. These results show that the knee extensor activity was significantly reduced from baseline preprogrammed stimulation levels with the implementation of a SCKM- and sensor-based control of FNS.
KNEE RANGE OF MOTION
A typical knee range of motion is shown in Figure 5 for FNS-only, HNP0, and HNP walking. The swing phase duration was consistent across conditions as defined by the preprogrammed stimulation pattern; however, the stance phase durations for the HNP0 and HNP were longer than those of FNS-only walking because of the manual triggering. There was a significant difference in the left but not the right (left, p < 0.001; right, p = 0.355) average maximum knee flexion with the HNP (left, 28° ± 4°; right, 36° ± 5°) relative to the HNP0 walking (left, 20° ± 4°; right, 34° ± 4°). Furthermore, no stimulus interpulse interval modulation from the FNS controller was observed, indicating that the baseline stimulus was consistently able to drive the knee into full extension at the end of swing before heel strike.
The average maximum knee flexion during FNS-only walking, shown in Figure 6D (left, 64° ± 5°; right, 83° ± 4°), was significantly greater than with the HNP or HNP0 (p < 0.001), as shown in Figure 6C, even though identical stimulation patterns were used. The influence of reciprocal hip coupling on knee flexion was considered by examining the average knee angle relative to the average thigh orientation as defined in Figure 6A during a stride for FNS only, HNP0, and HNP. As seen in Figure 6B, before the stimulation of the knee extensors (thick line) to extend the knee into the latter part of swing, knee flexion increases with thigh orientation. At 74° ± 11° (left) and 73° ± 4° (right), the average maximum hip flexion angles of the FNS-only case were significantly greater than the 18° ± 3° (left) and 23° ± 3° (right) of hip flexion observed with the HNP (left, p < 0.001; right, p < 0.001), where the hip reciprocator of the IRGO limited the hip motion. Thus, the anterior thigh orientation during HNP0 and HNP walking is less than with the FNS-only walking. This indicates that the knee flexion during swing is a function of thigh orientation regulated by the reciprocator of the IRGO.
This study examined the effect of an HNP combining a modified IRGO with a pair of hydraulic SCKMs27 and a sensor-based FNS controller on the upper limb loading, stimulation duty cycle, and knee range of motion. Feedback of the constraint state of each SCKM allowed for the deactivation of FNS to the knee extensor muscles when no knee extension was required. In addition, with feedback signals from sensors mounted on the orthosis, stimulation to the knee extensor muscles was modulated to reach the desired endpoints of the knee trajectories. The stimuli to the knee extensor muscles were exclusively targeted for modulation by the controller because they are primarily responsible for supporting body weight and their fatigue can be detrimental in maintaining FNS-only gait in persons with paraplegia.
The HNP combining an orthosis and FNS system for walking in paraplegia was demonstrated to have reduced the user’s upper limb loading relative to the use of either system alone. However, upper limb forces were significantly higher with HNP as compared with HNP0. The subject practiced and was comfortable walking with FNS only with knees locked in full extension during stance, which was provided by the knee extensor stimulation in the HNP0 system. In the HNP, knee was controlled with SCKM, which allowed about 13° of compliance. This may have given the subject the unfamiliar sensation of falling and prompted him to put extra weight on his arms.
In comparing the HNP to the FNS-only walking, the mean upper limb forces needed for balance and support were reduced by 36%, whereas relative to walking with the IRGO without FNS, both mean and average maximum upper limb forces were reduced by 17% and 30%, respectively. With FNS-only walking, the average forward trunk lean was 16° greater than with HNP. This suggests that when walking with FNS only, the upper limb loads applied to the walking aid were primarily for maintaining trunk posture in a leaning orientation. The HNP reduced the upper limb loading through the actions of the hip reciprocator of the IRGO, which provided anterior/posterior as well as mediolateral trunk support. Conversely, in IRGO-only walking, the upper limbs are used for hiking the hip during swing to attain sufficient foot-to-ground clearance and for moving the body forward by means of the hip reciprocator. The HNP reduced upper limb forces by providing knee flexion during swing and relying on lower limb muscles driven by FNS to propel the user forward. For IRGO-only walking, the contribution of the upper limb forces for propulsion as opposed to facilitating adequate foot-to-ground clearance during swing remains to be determined. However, significantly less trunk movement was necessary to facilitate stepping with HNP while applying lower upper limb forces on the walker.
At a mean walking speed of 0.14 m/s with the HNP, there was, on average, a 68% reduction in knee extensor stimulation duty cycle relative to FNS baseline stimulation. This allowed the knee extensors to rest for 74% of the stride duration. Manually triggered FNS stepping reduced cadence and increased the duration of double stance of each stride relative to free-cycling FNS-only stepping. For the HNP, this prolonged the period in which the stimulation was turned off relative to the period in which the stimulation was turned on by the FNS controller. Thus, it is expected that reductions in stimulation duty cycle by the FNS controller will be less with automatically triggered FNS stepping and increased gait speed than with manual triggering at lower walking speed.
The knee flexion driven by FNS while walking with the SCKM-based HNP was slightly greater and more symmetric than that with a conventional IRGO with unlocked knee joints driven by FNS (HNP0). In addition, the SCKM did not impede the FNS-driven knee extension at the end of swing in preparation for heel strike. This indicates that the additional mass and frictional/viscous effects of the SCKM did not significantly influence knee angle trajectory. However, the knee flexion during swing in HNP walking was substantially less than that observed in FNS-only walking even though identical stimulation patterns were implemented. This indicates that the knee flexion torque generated by the FNS was not a primary factor in contributing to the reduced knee motion. Results suggest that the reduced hip flexion due to the 1:1 reciprocal hip flexion-extension coupling by the IRGO reciprocator of the HNP may have impaired knee flexion over FNS-only walking. Preliminary HNP walking trials were conducted with the hip reciprocator disengaged, thereby uncoupling the hips. Increased knee flexion was observed at the expense of larger upper limb forces needed for trunk support relative to the HNP with reciprocally coupled hips. Furthermore, observed increases in hip and knee flexion angular velocities with the hip reciprocator disengaged indicate that the reciprocal hip coupling may significantly affect the inertial properties of the lower limb segments during swing. Future testing is needed to substantiate the influence of the hip reciprocator on knee dynamics during gait.
The reductions in upper limb loading and muscle stimulation observed with application of the SCKM and FNS controller relative to the existing assistive devices for walking in paraplegia have the potential of delaying the onset of fatigue in the upper limbs and knee extensor muscles, respectively. Thus, the combination of the SCKM and FNS controller may prolong walking durations and provide their users with the capability to walk clinically relevant distances. However, this needs to be verified in future testing of walking distance, elapsed time, and metabolic energy when the HNP becomes untethered from the laboratory and without the delay of manually triggered stepping.
1. Stallard J, Major RE, Poiner R, et al.. Engineering design considerations of the ORLAU Parawalker and FES hybrid system. Eng Med 1986; 15 (3): 123–129.
2. McClelland M, Andrews BJ, Patrick JH, et al.. Augmentation of the Oswestry Parawalker Orthosis by means of surface electrical stimulation: gait analysis of three patients. Paraplegia 1987; 25: 32–38.
3. Solomonow M, Baratta R, Hirokawa S, et al.. The RGO generation II: muscle stimulation powered orthosis as a practical walking system for thoracic paraplegics. Orthopedics 1989; 12: 1309–1315.
4. Nene AV, Jennings SJ. Hybrid paraplegic locomotion with the Parawalker using intramuscular stimulation: a single subject study. Paraplegia 1989; 27: 125–132.
5. Nene AV, Patrick JH. Energy cost of paraplegic locomotion using the Parawalker—electrical stimulation “hybrid” orthosis. Arch Phys Med Rehabil 1990; 71: 116–120.
6. Isakov E, Douglas R, Berns P. Ambulation using the reciprocating gait orthosis and functional electrical stimulation. Paraplegia 1992; 30: 239–245.
7. Yang L, Granat MH, Paul JP, et al.. Further development of hybrid functional electrical stimulation orthoses. Spinal Cord 1996; 34 (10): 611–614.
8. Solomonow M, Aguilar E, Reisin E, et al.. Reciprocating gait orthosis powered with electrical muscle stimulation (RGO II). Part I: performance evaluation of 70 paraplegic patients. Orthopedics 1997; 20 (4): 315–324.
9. Marsolais EB, Kobetic R, Polando G, et al.. The Case Western Reserve University hybrid gait orthosis. J Spinal Cord Med 2000; 23 (2): 100–108.
10. Kobetic R, Marsolais EB, Triolo RJ, et al.. Development of a hybrid gait orthosis: a case report. J Spinal Cord Med 2003; 26 (3): 254–258.
11. Rose GK. The principles and practice of hip guidance articulations. Prosthet Orthot Int 1979; 3: 37–43.
12. Douglas R, Larson PF, D’Ambrosia R, et al.. The LSU reciprocating gait orthosis. Orthopedics 1983; 6: 834–839.
13. Jefferson RJ, Whittle MW. Performance of three walking orthosis for the paralyzed: a case study using gait analysis. Prosthet Orthot Int 1990; 14: 103–110.
14. Motloch WM. Principles of orthotic management for child and adult paraplegia and clinical experience with the isocentric RGO. In: Proceedings of the 7th World Congress of the International Society for Prosthetics and Orthotics
. International Society for Prosthetics and Orthotics (ISPO): Copenhagen, Denmark; 1992:28.
15. Tashman S, Zajac FE, Perkash I. Modeling and simulation of paraplegic ambulation in a reciprocating gait orthosis. J Biomech Eng 1995; 117: 300–308.
16. Hirokawa S, Grimm M, Le T, et al.. Energy consumption in paraplegic ambulation using the reciprocating gait orthosis and electrical stimulation of the thigh muscles. Arch Phys Med Rehabil 1990; 71: 687–694.
17. Sykes L, Edwards J, Powell ES, et al.. The reciprocating gait orthosis: long-term usage patterns. Arch Phys Med Rehabil 1995; 76: 779–783.
18. Kralj A, Bajd T, Turk R. Electrical stimulation providing functional use of paraplegic patient muscles. Med Prog Technol 1980; 7: 3–9.
19. Bajd T, Kralj A, Turk R, et al.. The use of a four-channel electrical stimulator as an ambulatory aid for paraplegic patients. Phys Ther 1983; 63 (7): 1116–1120.
20. Marsolais EB, Kobetic R. Functional electrical stimulation for walking in paraplegia. J Bone Joint Surg 1987; 69A: 728–733.
21. Marsolais EB, Kobetic R. Development of a practical electrical stimulation system for restoring gait in the paralyzed patient. Clin Orthop 1988; 233: 64–74.
22. Gallien P, Brissot R, Eyssette M, et al.. Restoration of gait by functional electrical stimulation for spinal cord injured patients. Paraplegia 1995; 33 (11): 660–664.
23. Kobetic R, Triolo RJ, Marsolais EB. Muscle selection and walking performance of multichannel FES systems for ambulation in paraplegia. IEEE Trans Rehab Eng 1997; 5 (1): 23–29.
24. Kobetic R, Triolo RJ, Uhlir JP, et al.. Implanted functional electrical stimulation system for mobility in paraplegia: a follow-up case report. IEEE Trans Rehabil Eng 1999; 7 (4): 390–398.
25. Gregory CM, Bickel CS. Recruitment patterns in human skeletal muscle during electrical stimulation. Phys Ther 2005; 85 (4): 358–364.
26. Goldfarb M, Korkowski K, Harrold B, et al.. Preliminary evaluation of a controlled-brake orthosis for FES-aided gait. IEEE Trans Neural Syst Rehabil Eng 2003; 11 (3): 241–248.
27. To CS, Kobetic R, Bulea TC, et al.. Stance control knee mechanism for lower-limb support in hybrid neuroprosthesis
. J Rehabil Res Dev 2011; 48 (7): 839–850.
28. Nandurkar S, Marsolais EB, Kobetic R. Percutaneous implantation of iliopsoas for functional neuromuscular stimulation. Clin Orthop 2001; 389: 210–217.
29. Kobetic R, Marsolais EB. Synthesis of paraplegic gait with multichannel functional neuromuscular stimulation. IEEE Trans Rehabil Eng 1994; 2 (2): 66–79.
30. Kobetic R, To CS, Schnellenberger JR, et al.. Development of hybrid orthosis for standing, walking, and stair climbing after spinal cord injury. J Rehabil Res Dev 2009; 46 (3): 447–462.
31. Mortimer JT, Kaufman D, Roessmann U. Intramuscular electrical stimulation: tissue damage. Ann Biomed Eng 1980; 8: 235–244.
32. Caroll SC, Triolo RJ, Chizeck HJ, et al.. Tetanic response of electrically stimulated paralyzed muscle at varying interpulse intervals. IEEE Trans Biomed Eng 1989; 36 (7): 644–653.
33. To SC. Closed-Loop Control and Variable Constraint Mechanisms of a Hybrid Neuroprosthesis to Restore Gait After Spinal Cord Injury
[dissertation]. Cleveland, OH: Department of Biomedical Engineering, Case Western Reserve University; 2010.
34. Kagaya K, Shimada Y, Sato K, et al.. An electrical knee lock system for functional electrical stimulation. Arch Phys Med Rehabil 1996; 77: 870–873.
35. Trier S, Vrabec T, Weisgarber J. Using functional electrical stimulation to restore movement to individuals with neuromuscular disabilities. MATLAB Digest: Academic Edition
36. Davis RB III, Ounpuu A, Tyburski D, et al.. A gait analysis data collection and reduction technique. Hum Move Sci 1991; 10: 575–587.
37. Audu ML, To CS, Kobetic R, et al.. Gait evaluation of a novel hip constraint orthosis with implication for walking in paraplegia. IEEE Trans Neural Syst Rehabil Eng 2010; 18 (6): 610–618.