Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Dec 1.
Published in final edited form as: IEEE Trans Neural Syst Rehabil Eng. 2019 Oct 29;27(12):2353–2360. doi: 10.1109/TNSRE.2019.2950226

Use of pelvic corrective force with visual feedback improves paretic leg muscle activities and gait performance after stroke

Chao-Jung Hsu 1, Janis Kim 1, Elliot J Roth 1, William Z Rymer 1, Wu Ming 1
PMCID: PMC6939618  NIHMSID: NIHMS1546535  PMID: 31675335

Abstract

The purpose of this study was to examine the effects of combined pelvic corrective force and visual feedback during treadmill walking on paretic leg muscle activity and gait characteristics in individuals with post-stroke hemiparesis. Fifteen chronic stroke participants completed visual feedback only and combined pelvic corrective force and visual feedback conditions during treadmill walking. Each condition included: 1-minute baseline, 7-minute training with visual feedback only or additional pelvic corrective force, 1-minute post training, 1- minute standing break, and another 5-minute training. EMGs from the paretic leg muscles and step length were measured. Overground walking was evaluated before treadmill walking, immediately and 10 minutes after treadmill walking. Greater increases in integrated EMG of all muscles, except vastus medialis and tibialis anterior, were observed with the application of additional pelvic corrective force compared to visual feedback only during treadmill walking. Overground walking speed significantly increased after treadmill training with combined pelvic correction force and visual feedback, but was not significant for the visual feedback only condition. Voluntary weight shifting with additional pelvic corrective force enhanced paretic leg muscle activities and improved gait characteristics during walking. Individuals with post-stroke hemiparesis could adapt feedforward control and generalize the adaptation to overground walking.

Keywords: EMG, locomotion, pelvic corrective force, stroke, visual feedback

I. Introduction

ALTHOUGH the majority of individuals with post-stroke hemiparesis could ultimately regain locomotor function after rehabilitation, many of them show deficits in gait patterns, including reduced paretic leg muscle activity [1] and decreased step length [2]. These gait deficits after stroke are associated with reduced weight bearing toward the paretic leg [3], which may be a consequence of a “learned nonuse”[4] and could be overcome by forced use of the paretic leg [5]. The forced use of the paretic leg has been advocated in gait rehabilitation to improve paretic leg motor function [6], but there are limited evidences that support this intervention paradigm for a better locomotor performance.

Constraint-induced movement therapy (CIMT) is a promising intervention to overcome the learned nonuse of the paretic arm [7]. It has been a challenge to apply CIMT to locomotor training because the feasibility of immobilizing the non-paretic leg during walking. A recent study that used a shoe insole to promote body weight shift towards the paretic leg showed improvements in symmetry of spatiotemporal gait characteristics [8], but gait velocity was not changed by the use of insole. Our previous study applied a corrective force to the pelvis during treadmill walking and showed increases in symmetry of pelvis displacement and muscle activity of the paretic leg [9], suggesting that applying pelvic correction force may effectively induce forced use of the paretic leg. However, it remains unclear whether the motor skills that were obtained from treadmill walking could be transferred to overground walking.

Visual feedback is another approach that has been used to facilitate weight bearing toward the paretic leg in stroke rehabilitation [10]. Providing visual feedback during treadmill walking improved gait velocity, spatiotemporal gait characteristics and mobility [11], [12]. In addition, the use of visual feedback to alter gait-related outcomes may encourage an active engagement for patients [13], which is more effective than passive training in eliciting performance improvements. While providing visual feedback could encourage patients with post-stroke hemiparesis to voluntarily shift body weight [10], many patients still showed insufficient weight shifting towards the paretic side due to weakness of the paretic leg. We postulated that the combination of pelvic corrective force and visual feedback during treadmill walking would lead to improved weight shifting toward the paretic leg and result in improvement in locomotor performance.

The aim of the study was to examine the effects of combined pelvic corrective force and visual feedback during treadmill walking on paretic leg muscle activity and gait characteristics in individuals with post-stroke hemiparesis. Based on available literature, we hypothesized that compared to visual feedback only, combined pelvic corrective force and visual feedback would induce greater enhancement in paretic leg muscle activity during treadmill walking. In addition, we hypothesized that gait characteristics during overground walking would be improved after treadmill training, suggesting a potential transfer of motor skills from treadmill to overground walking.

II. Methods

A. Participants

Fifteen participants with chronic (> 6 months) post-stroke hemiparesis were recruited in this study. The Northwestern University Institutional Review Board approved the study protocol and all participants provided written informed consent to participate in this study. Patients were recruited from an existing database of volunteers, and through flyers and word of mouth. Inclusion criteria for the patients were: (1) age 21–75 years, (2) unilateral, supratentorial, ischemic or hemorrhagic stroke confirmed with radiography, (3) no prior history of stroke before the reference stroke, (4) independent ambulation with/without the use of assistive device or below knee orthoses, (5) self-selected walking speed ≤ 0.80 m/s. Exclusion criteria for the patients were: (1) brainstem or cerebellar stroke, (2) a score on the Mini Mental Status examination < 24 [14], (3) other neurological conditions, cardiorespiratory/metabolic disorders, or orthopedic conditions affecting ambulation ability, (4) botox injection within 6 months of study enrollment visit.

B. Apparatus

A customized cable-driven robotic system based over a treadmill was used in this study [15]. The robotic system consisted of 2 nylon-coated stainless-steel cables, driven by 2 motors (AKM 33H, Kollmorgen, Radford, VA). The cables were affixed to a custom waist belt that was strapped to participants’ pelvis to provide a corrective force in the lateral direction toward the paretic side while participants walked on the treadmill. Two custom designed 3-D position sensors were attached at participants’ ankles to record ankle position signals. The position sensor consists of one detector bar and two U-joints, which were attached to the two ends of the detector bar. Two rotational potentiometers (P2201, Novotechnik, Southborough, MA) and one linear potentiometer (SP-2, Celesco, Chatsworth, CA) were used to measure the rotational movement of the bar in the anterior-posterior and medial-lateral direction, and the length of the telescoping bar. The position sensor was attached at the subject’s ankle through a strap and a U-joint. A custom-written LabVIEW program was used to command corrective force signals to the motors.

Visual feedback was provided through a TV screen that was placed in front of the treadmill. A vertical bar was displayed on the screen with its height was changed to reflect the magnitude of participants’ weight bearing on the paretic leg in real-time (Fig. 1). Weight bearing on the paretic leg was monitored by using a mobile SmartStep device (version 2.20, Andante Medical Devices). The SmartStep is a portable biofeedback system that consists of a pneumatic insole measuring the vertical force under the foot at a frequency of 40 Hz. Vertical force data were transmitted to a computer through wireless technology and displayed on the TV screen.

Fig. 1.

Fig. 1.

Open in a new tab

Experimental setup with the locations of position sensors and computer monitor for visual feedback.

An instrumented gait mat (GaitRite, CIR Systems, Clifton, NJ) was used to measure overground gait speed and step length. The gait mat contains a total of 16,128 sensors and the gait mat data were sampled at 120 Hz.

C. Experimental protocol

Each participant completed 2 testing conditions, which included visual feedback only, and combined pelvic corrective force and visual feedback conditions. A 10-minute sitting break was inserted between 2 testing conditions. The order of 2 testing conditions was randomized across participants. The testing protocol consisted of both treadmill and overground walking sessions. During the treadmill walking session, participants walked at self-selected comfortable speed on the treadmill (Woodway USA, Waukesha, WI) with an overhead harness that was used for protection only (no body weight support was provided). The self-selected comfortable speed was determined at the beginning of the data collection and remained the same for the 2 testing conditions. Each participant also walked at his/her maximum walking speed for 30 strides before starting the first testing session for the purpose of EMG normalization. As shown in Fig. 2, each testing condition included 5 sessions: 1-minute baseline, 7-minute training with visual feedback only or combined pelvic corrective force and visual feedback (adaptation), 1-minute post training (post adaptation), 1-minute standing break, and 5-minute training with visual feedback only or combined pelvic corrective force and visual feedback. The magnitude of the pelvic corrective force was set at ~9% of participants’ body weight, which was adjusted if participants could not tolerate. The pelvic corrective force was triggered at the timing of heel strike of the paretic leg and lasted for 400ms. The heel strike was determined using the ankle position signals. A 5-Newton pretension force was applied to the cables in order to prevent cable slacking. For visual feedback, the target weight bearing was set at 120% of weight bearing of the paretic leg during overground walking (which was determine based on our pilot study) and was displayed as a horizontal line on the TV screen. Participants were instructed to shift more body weight onto the paretic leg. When the body weight shifted on the paretic leg exceeded the target weight bearing, the vertical bar would go above the horizontal line and a beep sound would be generated by the computer. Participants were allowed to hold onto the front handrail for safety. Overground walking was evaluated before treadmill walking, immediately after treadmill walking, and 10 minutes after treadmill walking. During overground walking session, participants walked at their self-selected comfortable speed for 3 trials (i.e., no any familiarization trials). Results reported were the average of these 3 trials.

Fig. 2.

Fig. 2.

Open in a new tab

Experimental protocol of 2 testing conditions.

D. Measurements and data analysis

Surface electrodes (Delsys DE 2.1, Delsys Inc., Boston, MA) were used to record the electromyograms (EMG) from paretic leg muscles: abductors (gluteus medius, ABD), adductor longus (ADD), medial hamstrings (MH), medial gastrocnemius (MG), soleus (SOL), rectus femoris (RF), vastus medialis (VM), and tibialis anterior (TA). EMG signals were amplified with gain of 1,000 and band-pass filtered of 20–450 Hz in hardware, and then sampled at 500 Hz via a 12-bit analog-to-digital converter (National Instruments, Austin, TX) on a PC running custom-written LabVIEW program.

EMG signals were analyzed using a custom software written in MATLAB. The EMG data was first low-pass filtered at 250 Hz, high-pass filtered at 10 Hz and band-stop filtered at 55–65 Hz using a second order Butterworth filter. The EMG data were then rectified and smoothed using a second order Butterworth filter with a low-pass cutoff frequency of 20 Hz. The smoothed EMG signals were segmented into gait cycles based on ankle positions. The smoothed EMG signals were interpolated, resampled, normalized to percentage of the gait cycle, and normalized to the peak value of EMG signals of each muscle when participants walked with their maximum walking speed. The integral of the EMG activity was calculated during the stance phase of gait of each muscle. In addition, for MG and SOL, the integrals of the EMG during the late stance phase were also calculated. The integrated EMG signals were then averaged for the last 5 strides at baseline (BASELINE), first 5 strides at adaptation (EARLY ADAPTATION), and first 2 strides at post adaptation (POST ADAPTATION). The means of integrated EMG of each muscle during baseline were also subtracted from the integrated EMG during adaptation period to compare the changes in EMG activity between two testing conditions. During the treadmill walking session, ankle position signals were recorded using a data acquisition card (National Instruments, Austin, TX) at a sample rate of 500 Hz. A custom-written LabVIEW program was used for data acquisition. Ankle position signals were analyzed using a custom-written MATLAB program (The Mathworks, Natick, MA). Step length was calculated as the anteroposterior distance between the two legs’ ankle positions at initial contact [2]. Step length was averaged for BASELINE, EARLY ADAPTATION, and POST ADAPTATION. For step length during treadmill and overground walking, step length asymmetry was quantified using the step length ratio, which was defined as the paretic step length divided by the non-paretic step length [17]. Weight bearing on the paretic leg data were collected using the SmartStep software, and the peak vertical force signals were averaged for the time periods of BASELINE, EARLY ADAPTATION, LATE ADAPTATION, and POST ADAPTATION (1 minute for each period). During the overground session, gait speed and step length were calculated using the GaitRite software. Step length ratio was also calculated.

E. Statistical analysis

Repeated measure ANOVAs were used to compare vertical force, integrated EMG of the paretic leg, overground walking speed, and step length ratio across time periods. A further post-hoc analysis was conducted if ANOVA yielded significant results. An alpha level of < 0.05 was set for significance. Paired t-test was used to compare the changes in integrated EMG between two testing conditions during the periods of EARLY ADAPTATION and LATE ADAPTATION. Bonferroni corrections were applied when appropriate.

III. Results

Demographics characteristics of participants are presented in Table 1. The average year after stroke was 7.9 ± 5.2 years. The average self-selected speed on the treadmill was 0.52 ± 0.12 m/s.

TABLE I.

Demographic information for the study participants

Number Sex Age Weight (kg) Post injury (yr) Assistive device Self-selected comfortable speed (m/s) Pelvic corrective force (N)
S1 F 64 88.5 19 AFO 0.37 52
S2 F 64 68.0 5 SPC/AFO 0.54 58
S3 M 59 90.7 5 AFO 0.56 72
S4 F 46 63.5 3 SPC/AFO 0.53 38
S5 M 54 85.7 4 AFO 0.45 60
S6 F 48 95.2 11 AFO 0.51 84
S7 M 56 81.7 10 AFO 0.49 80
S8 F 58 77.1 4 AFO 0.79 72
S9 M 68 78.5 5 SPC/AFO 0.43 50
S10 M 48 68.0 13 Carbon PLS 0.69 60
S11 F 56 68.0 7 AFO 0.57 60
S12 M 65 69.9 3 SPC/AFO 0.39 46
S13 F 58 72.1 8 Carbon PLS 0.65 86
S14 F 50 84.8 4 AFO 0.40 50
S15 M 50 85.3 18 AFO 0.46 70
Open in a new tab

AFO = ankle foot orthosis; SPC = single point cane; Carbon PLS = carbon fiber ankle foot orthotic posterior leaf spring.

A. Weight bearing on the paretic leg

Weight bearing on the paretic leg from 15 participates were analyzed. For the combined pelvic corrective force and visual feedback condition, as shown in Fig. 3, weight bearing on the paretic leg significantly increased (F2,14 = 5.19, p = 0.012). Post-hoc tests indicated significant increases in weight bearing on the paretic leg from 56.9 ± 16.1 kg BASELINE to 60.0 ± 15.4 kg, p = 0.028, at EARLY ADAPTATION, and to 62.5 ± 15.2 kg, p = 0.022, at LATE ADAPTATION. For the visual feedback only condition, weight bearing on the paretic leg also had significant change (F2,14 = 6.39, p = 0.005). Post-hoc tests indicated that weight bearing had no significant change at EARLY ADAPTATION, p = 0.61, but significantly increased from 56.7 ± 15.5 kg BASELINE to 60.8 ± 15.0 kg, p = 0.001, at LATE ADAPTATION. During the POST ADAPTATION period, weight bearing on the paretic leg were significantly greater than BASELINE, for the combined condition (i.e., 60.0 ± 15.8 kg, p = 0.003), and visual feedback only conditions (59.1 ± 16.0 kg, p = 0.005), respectively.

Fig. 3.

Fig. 3.

Open in a new tab

Weight bearing on the paretic leg during baseline (BA), early adaptation (EAD), and late adaptation (LAD) periods for both the visual feedback only (VF only) and combined pelvic force and visual feedback conditions (VF + Pelvis). Data shown are mean and standard error across 15 subjects. ‘*’ indicates significant difference.

B. Integrated EMG

EMG data from 15 participants were analyzed. EMG data of VM muscle from 1 participant were excluded due to large artifacts. Muscle activation patterns of each muscle from a representative participant for the visual feedback only condition and combined condition during BASELINE, EARLY ADAPTATION, LATE ADAPTATION, and POST ADAPTATION are shown in Fig. 4 and Fig. 5, respectively. For the visual feedback only condition, muscle activities in MH, VM, and SOL had modest changes during the EARLY ADAPTATION, but were higher during the LATE ADAPTATION (Fig. 4). For the combined condition, higher muscle activities in MG, SOL, and VM were observed during EARLY ADAPTATION (Fig. 5).

Fig. 4.

Fig. 4.

Open in a new tab

Muscle activation patterns of medial hamstrings, vastus medialis, soleus and tibialis anterior in visual feedback only condition from a representative participant. Data were averaged across 5 gait cycles during BASELINE, EARLY ADAPTATION, LATE ADAPTATION and POST ADAPTATION, and normalized to % gait cycle.

Fig. 5.

Fig. 5.

Open in a new tab

Muscle activation patterns of medial hamstrings, vastus medialis, soleus and tibialis anterior in visual feedback only condition from a representative participant. Data were averaged across 5 gait cycles during BASELINE, EARLY ADAPTATION, LATE ADAPTATION and POST ADAPTATION, and normalized to % gait cycle.

Group average of integrated EMG indicated that the magnitude of most muscles of the paretic leg enhanced for the combined condition, but had modest change for the visual feedback only condition during EARLY ADAPTATION, Fig. 6. Specifically, participants showed significantly greater increases in integrated EMG of all muscles, except TA and VM during EARLY ADAPTATION for the combined condition compared to the visual feedback only condition (ABD, p = 0.007; ADD, p = 0.043 (n = 13, data from one subjects were excluded due to outlier, which was defined as a value that was more than three scaled median absolute deviations away from the median); MH, p < 0.001; MG, p = 0.003; SOL, p = 0.020 (n = 13); RF, p = 0.035; VM, p = 0.247; TA, p = 0.369, Fig. 7. Changes in integrated EMG during the LATE ADAPTATION showed significant difference for MH (p = 0.028, n = 12) and ABD (p = 0.005, n = 13), but were not significant for other muscles (p > 0.05), Fig. 7.

Fig. 6.

Fig. 6.

Open in a new tab

Step-by-step changes in integrated EMG of the paretic leg during treadmill walking for the conditions with pelvic correction force and visual feedback (pelvis + VF) and visual feedback only (VF). Data shown are average across 14 subjects post stroke (data from one subject were excluded due to large artifacts). The step numbers were different across subjects during adaptation period because the treadmill speeds were different. Thus, the data from the first 120 steps and last 20 steps during the adaptation period were used to calculate the average of integrated EMG. The mean of the EMG during baseline were subtracted from the integrated EMGs during the course of walking.

Fig. 7.

Fig. 7.

Open in a new tab

Average of the changes in integrated EMG during early adaptation (EAD) and late adaptation (LAD) adaptation periods. * indicates significant difference. Data shown are mean and standard error.

Providing pelvic corrective force and visual feedback had significant impact on integrated EMGs of RF (F12,2 = 3.24, p = 0.057), MH (F13,2 = 5.28, p =0.012), and ABD (F13,2 = 3.86, p = 0.034). Post-hoc analysis indicated that integrated EMG of MH significantly increased from 0.85 ± 0.26 BASELINE to 1.05 ± 0.23 at EARLY ADATPATION, increased 31.5%, p = 0.024, and to 1.05 ± 0.32 at POST ADAPTATION, 29.9%, p = 0.024, Fig. 8. Integrated EMG of RF significantly increased from 0.87 ± 0.20 at BASELINE to 1.00 ± 0.19 at EARLY ADATPATION, 17.5%, p = 0.046, and was 0.95 ± 0.27 at POST ADAPTATION, 10.3%, but this was not significant, p = 0.34. Integrated EMG of ABD significantly increased from 1.00 ± 0.18 BASELINE to 1.20 ± 0.36 at POST ADAPTATION, 21.7%, p = 0.041, and tended to increase to 1.17 ± 0.18 at EARLY ADATPATION, 19.0%, p = 0.09 > 0.05. Providing pelvic corrective force and visual feedback had no significant impact on integrated EMGs of other muscles (p > 0.05). In addition, subgroup analysis indicated that integrated EMG of MG during the late stance phase significantly increased from 0.24 ± 0.11 at BASELINE to 0.32 ± 0.17, at EARLY ADAPTATION, 31.8%, p = 0.047, although had no significant change at POST ADAPTATION, p = 0.40. The integrated EMG of SOL during late stance phase had no significant change (p = 0.28).

Fig. 8.

Fig. 8.

Open in a new tab

Average of integrated EMG during baseline, early adaptation (EAD), and post adaptation (PAD) for the visual feedback only (VF) and pelvic correction force and visual feedback (pelvis + VF) conditions. Error bars: ± standard error. * Significant difference between periods (p<0.05).

Group average results indicated that the impact of providing visual feedback only on the integrated EMG during stance was not significant for all muscles (p > 0.05). In addition, subgroup analysis indicated that the integrated EMG of MG and SOL during the late stance phase tended to increase from 0.28 ± 0.17 at BASELINE to 0.40 ± 0.35 at POST ADAPTATION, p = 0.068, for MG, and from 0.35 ± 0.16 at BASELINE to 0.42 ± 0.15 at POST ADAPTATION, p = 0.073, for SOL, but these changes were not significant.

C. Step length asymmetry and overground gait speed

During the treadmill walking session, for the combined condition, step length ratio increased from 0.89 ± 0.24 at BASELINE to 1.04 ± 0.28, p = 0.004, at EARLY ADAPTATIO, and decreased from EARLY ADAPTATION to 0.92 ± 0.31, p = 0.03, at LATE ADAPTATION, and to 0.86 ± 0.36, p = 0.016 at POST ADAPTATION. For the visual feedback only condition, step length ratio had no significant change (p > 0.05).

Treadmill walking with the combined pelvic corrective force and visual feedback had significant impact on overground walking speed (F14,2 = 5.12, p = 0.013). Post-hoc analysis indicated that gait speed significantly increased from 0.73 ± 0.17 m/s before treadmill walking to 0.78 ± 0.17 m/s, immediately after treadmill walking (p = 0.030), and to 0.78 ± 0.16m/s at 10 mins after treadmill walking, p = 0.022). Step length ratio also had significant change (F14,2 = 3.62, p = 0.04). Post-hoc analysis indicated that step length ratio tended to increase from 1.14 ± 0.18 before treadmill walking to 1.19 ± 0.20, immediately after treadmill walking (p = 0.068), and returned back to 1.14 ± 0.15 at 10 mins after treadmill walking (p = 1.00).

Treadmill walking with visual feedback only had no significant impact on overground walking speed (p = 0.070 > 0.05), although walking speed tended to increase from 0.74 ± 0.17 m/s before treadmill walking to 0.76 ± 0.16 m/s, immediately after treadmill walking, and to 0.78 ± 0.17m/s at 10 mins after treadmill walking). Providing visual feedback only had significant impact on step length ratio, F14,2 = 4.72, p = 0.017. Post-hoc analysis indicated that step length ratio significantly increased from 1.11 ± 0.15 before treadmill walking to 1.18 ± 0.18 (i.e., more asymmetrical), p = 0.026, immediately after treadmill walking, and to 1.17 ± 0.16, p = 0. 044. Changes in overground walking speed after treadmill walking with combined pelvic correction force and visual feedback, and with visual feedback only was not significantly different (p = 0.33).

IV. Discussion

In this study, we examined the effects of combined pelvic corrective force and visual feedback during treadmill walking on paretic leg muscle activity and gait characteristics in individuals with post-stroke hemiparesis. As hypothesized, greater paretic leg muscle activities were observed with the application of combined pelvic corrective force and visual feedback compared to visual feedback only. Increased muscle activity of ABD and MH were retained following the removal of the combined pelvic corrective force and visual feedback. Stroke patients shifted more body weight on the paretic leg with the application of combined pelvic corrective force and visual feedback during treadmill walking. The training effect on step length was also found. Step length symmetry improved during EARLY ADAPTATION with the application of combined pelvic correction force and visual feedback. Walking speed significantly increased during overground walking after treadmill training with the combined pelvic corrective force and visual feedback. Our results suggest that providing pelvic corrective force and visual feedback during treadmill walking could potentially promote use of the paretic leg. The training effect obtained during treadmill walking was able to transfer to overground walking.

Individuals with post-stroke hemiparesis often demonstrate impaired ability to shift weight onto their paretic leg [18]-[20]. Visual feedback in our study provided concurrent feedback on weight bearing, which is the form of knowledge of performance. The benefit of knowledge of performance could be contributed to the cognitive effort induced by the feedback. While providing visual feedback about weight bearing of the paretic leg, our participants consciously shifted body weight onto the paretic leg during treadmill walking. With additional pelvic corrective force, our participants were forced to shift more weight onto the paretic leg. Both interventions increased loading onto the paretic leg, which may enhance load afferents inputs from the Golgi tendon organs in ankle extensor muscles and cutaneous afferents input from the sole of the foot [21], [22], resulting in enhanced leg extensor muscle activations. It is likely that increased muscle activities of gastrocnemius and soleus during the late stance phase also reinforced ankle push-off force, which assisted the leg swing further and thus increased step length.

A major cause of asymmetric weight shifting during walking after stroke is the lack of hip abductor activation. Hip abductor muscle weakness also affects the hip joint position during stance, which is an important factor affecting the recovery of independent gait [23]. The increased gluteus medius activity in our participants may improve the hip joint stability for single leg support of the paretic leg in both sagittal and frontal planes [25].

Increased paretic leg muscle activities lasted beyond the adaptation period, suggesting that repetitively forced use of the paretic leg during walking may promote the use of the paretic leg through a use-dependent motor learning mechanism [26]. The retention after the adaptation period indicated the predictive feedforward mechanism was involved. Feedforward mechanism anticipates expected perturbations and incorporates expected effects into subsequent steps. Temporary motor memory might have been formed after repetitively forced use of the paretic leg. In addition, repetitive movement in a particular pattern overtime may bias future movement in favor the particular pattern even when the external perturbation is removed [27]. Therefore, increased paretic leg muscle activities were still observed after the removal of pelvic corrective force and visual feedback. Our result is partially consistent with a recent stroke rehabilitation study that showed greater paretic anterior ground reaction force after the removal of real-time visual feedback [28]. The predictive locomotor adaptation is dependent on cerebellar integrity [29]. Research has reported that this type of locomotion adaptation is demonstrated in decerebrate cats [30], suggesting cerebrum may not be critical for the locomotion adaptation. As such, our intervention appears to be appropriate for stroke survivors with no cerebellar lesion.

Combined pelvic corrective force and visual feedback improved step length ratio during treadmill walking. Different mechanisms for the improvement may be involved due to the heterogeneous patterns of asymmetrical step length in our participants. For participants who walk with a shorter step length on the non-paretic leg, increased both RF and MH muscle activities improved the stability of the paretic leg during standing [25] that allowed participants to take a longer step on the non-paretic leg. In addition, increased loading onto the paretic leg might facilitate trunk progression during paretic stance, which could assist non-paretic leg swing forward [31] and thus increase non-paretic step length. For participants who walk with a shorter step length on the paretic leg, it is likely that increased MG and SOL muscle activities reinforced ankle propulsion force at late stance phase to facilitate leg swing further and thus increased paretic step length.

Our finding indicates that the effect of combined pelvic corrective force and visual feedback could be generalized from treadmill to overground walking. With the repetitive and task-specific practice of enhanced use of the paretic leg during treadmill walking, participants were able to transfer these motor skills to overground walking, which may be due to the overlapping of neural control circuities involved in both the treadmill and overground walking. As a result, overground walking speed increased after treadmill walking with combined pelvic correction force and visual feedback. Thus, this intervention paradigm is promising for motor skills acquisition and retention of lower extremity activities after stroke [32].

This study has limitations to consider when interpreting the results. First, all participants held onto the front handrail during treadmill walking. It is possible that the pulling force from participants’ arm could compromise the training effect. Unfortunate, we were not able to collect the force data from the arm using an instrumented handrail. Wearing AFO, particularly a rigid AFO, might diminish the impact of the pelvic corrective force and visual feedback on the ankle muscle activity of the paretic leg [33], although we still observed significant increases in ankle muscle activity. In particular, because all participants wore an AFO, the impacts should be consistent across participants. In addition, clinical measures, such as muscle tone, were not evaluated in this study. It is unknown whether our participants with different clinical presentations would respond differently to our interventions. Finally, the number of participants is small, which limits our ability to examine the training effects on participants with different locations of lesion and different levels of walking function. Future research with larger sample size is needed to expand the understand the involvement of neural substrates in order to apply the constraint induced movement therapy to improve the motor function of the paretic leg of stroke survivors with different locations of lesion.

V. Conclusion

Voluntary weight shifting paired with additional pelvic corrective force could enhance paretic leg muscle activities and improve gait characteristics during treadmill walking in individuals with post-stroke hemiparesis. In addition, individuals with post-stroke hemiparesis were able to adapt feedforward control and generalize the adaptation to overground walking. Results from this study could potentially be used to develop a constraint induced movement therapy paradigm for locomotor training. Future research is needed to determine whether our treatment protocol is effective for a broad spectrum of stroke survivors.

Acknowledgments

This work was supported by the National Institutes of Health, R01HD082216.

References

  • [1].Lamontagne A, Stephenson JL, and Fung J, “Physiological evaluation of gait disturbances post stroke,” Clinical Neurophysiology, vol. 118, no. 4, pp. 717–729, April 2007. [DOI] [PubMed] [Google Scholar]
  • [2].Patterson KK, Gage WH, Brooks D, Black SE, and McIlroy WE, “Changes in gait symmetry and velocity after stroke: a cross-sectional study from weeks to years after stroke,” Neurorehabil Neural Repair, vol. 24, no. 9, pp. 783–790, December 2010. [DOI] [PubMed] [Google Scholar]
  • [3].Hendrickson J, Patterson KK, Inness EL, McIlroy WE, and Mansfield A, “Relationship between asymmetry of quiet standing balance control and walking post-stroke,” Gait & Posture, vol. 39, no. 1, pp. 177–181, January 2014. [DOI] [PubMed] [Google Scholar]
  • [4].Kwakkel G, Veerbeek JM, van Wegen EEH, and Wolf SL, “Constraint-induced movement therapy after stroke,” Lancet Neurol, vol. 14, no. 2, pp. 224–234, February 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Taub E, Uswatte G, King DK, Morris D, Crago JE, and Chatterjee A, “A placebo-controlled trial of constraint-induced movement therapy for upper extremity after stroke,” Stroke, vol. 37, no. 4, pp. 1045–1049, April 2006. [DOI] [PubMed] [Google Scholar]
  • [6].Billinger SA, Gajewski BJ, Guo LX, and Kluding PM, “Single limb exercise induces femoral artery remodeling and improves blood flow in the hemiparetic leg poststroke,” Stroke, vol. 40, no. 9, pp. 3086–3090, September 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Wolf SL et al. , “Effect of constraint-induced movement therapy on upper extremity function 3 to 9 months after stroke: the EXCITE randomized clinical trial,” JAMA, vol. 296, no. 17, pp. 2095–2104, November 2006. [DOI] [PubMed] [Google Scholar]
  • [8].Ma CC, Rao N, Muthukrishnan S, and Aruin AS, “A textured insole improves gait symmetry in individuals with stroke,” Disabil Rehabil, vol. 40, no. 23, pp. 2798–2802, 2018. [DOI] [PubMed] [Google Scholar]
  • [9].Hsu CJ, Kim J, Tang R, Roth EJ, Rymer WZ, and Wu M, “Applying a pelvic corrective force induces forced use of the paretic leg and improves paretic leg EMG activities of individuals post-stroke during treadmill walking,” Clinical neurophysiology, vol. 128, no. 10, pp. 1915–1922, October 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].van Vliet PM and Wulf G, “Extrinsic feedback for motor learning after stroke: what is the evidence?,” Disabil Rehabil, vol. 28, no. 13–14, pp. 831–840, July 2006. [DOI] [PubMed] [Google Scholar]
  • [11].Drużbicki M et al. , “The efficacy of gait training using a body weight support treadmill and visual biofeedback in patients with subacute stroke: a randomized controlled trial,” BioMed Research International, 2018. [Online]. Available: https://www.hindawi.com/journals/bmri/2018/3812602/ [DOI] [PMC free article] [PubMed]
  • [12].Lewek MD, Feasel J, Wentz E, Brooks FP, and Whitton MC, “Use of visual and proprioceptive feedback to improve gait speed and spatiotemporal symmetry following chronic stroke: a case series,” Phys Ther, vol. 92, no. 5, pp. 748–756, May 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [13].van Gelder LMA, Barnes A, Wheat JS, and Heller BW, “The use of biofeedback for gait retraining: A mapping review,” Clin Biomech (Bristol, Avon), vol. 59, pp. 159–166, November 2018. [DOI] [PubMed] [Google Scholar]
  • [14].Folstein MF, Folstein SE, and McHugh PR, “‘Mini-mental state’. A practical method for grading the cognitive state of patients for the clinician,” J Psychiatr Res, vol. 12, no. 3, pp. 189–198, November 1975. [DOI] [PubMed] [Google Scholar]
  • [15].Wu M, Hornby TG, Landry JM, Roth H, and Schmit BD, “A cable-driven locomotor training system for restoration of gait in human SCI,” Gait & Posture, vol. 33, no. 2, pp. 256–260, February 2011. [DOI] [PubMed] [Google Scholar]
  • [16].Leech KA, Kinnaird CR, Holleran CL, Kahn J, and Hornby TG, “Effects of locomotor exercise intensity on gait performance in individuals with incomplete spinal cord injury,” Phys Ther, vol. 96, no. 12, pp. 1919–1929, December 2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].Balasubramanian CK, Bowden MG, Neptune RR, and Kautz SA, “Relationship between step length asymmetry and walking performance in subjects with chronic hemiparesis,” Arch Phys Med Rehabil, vol. 88, no. 1, pp. 43–49, January 2007 [DOI] [PubMed] [Google Scholar]
  • [18].Goldie PA, Matyas TA, Evans OM, Galea M, and Bach TM, “Maximum voluntary weight-bearing by the affected and unaffected legs in standing following stroke,” Clin Biomech (Bristol, Avon), vol. 11, no. 6, pp. 333–342, September 1996. [DOI] [PubMed] [Google Scholar]
  • [19].Pai YC, Rogers MW, Hedman LD, and Hanke TA, “Alterations in weight-transfer capabilities in adults with hemiparesis,” Phys Ther, vol. 74, no. 7, pp. 647–657–659, July 1994. [DOI] [PubMed] [Google Scholar]
  • [20].Dorsch S, Ada L, and Canning CG, “Lower limb strength is significantly impaired in all muscle groups in ambulatory people with chronic stroke: a cross-sectional study,” Arch Phys Med Rehabil, vol. 97, no. 4, pp. 522–527, April 2016. [DOI] [PubMed] [Google Scholar]
  • [21].Bastiaanse CM, Duysens J, and Dietz V, “Modulation of cutaneous reflexes by load receptor input during human walking,” Exp Brain Res, vol. 135, no. 2, pp. 189–198, November 2000. [DOI] [PubMed] [Google Scholar]
  • [22].Dietz V and Duysens J, “Significance of load receptor input during locomotion: a review,” Gait & Posture, vol. 11, no. 2, pp. 102–110, April 2000. [DOI] [PubMed] [Google Scholar]
  • [23].Pandy MG and Andriacchi TP, “Muscle and joint function in human locomotion,” Annu Rev Biomed Eng, vol. 12, pp. 401–433, August 2010. [DOI] [PubMed] [Google Scholar]
  • [24].Powers CM, “The influence of altered lower-extremity kinematics on patellofemoral joint dysfunction: a theoretical perspective,” J Orthop Sports Phys Ther, vol. 33, no. 11, pp. 639–646, November 2003. [DOI] [PubMed] [Google Scholar]
  • [25].De Visser E, Mulder T, Schreuder HW, Veth RP, and Duysens J, “Gait and electromyographic analysis of patients recovering after limbsaving surgery,” Clin Biomech (Bristol, Avon), vol. 15, no. 8, pp. 592–599, October 2000. [DOI] [PubMed] [Google Scholar]
  • [26].Mawase F, Uehara S, Bastian AJ, and Celnik P, “Motor learning enhances use-dependent plasticity,” J. Neurosci, vol. 37, no. 10, pp. 2673–2685, 08 2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Classen J, Liepert J, Wise SP, Hallett M, and Cohen LG, “Rapid plasticity of human cortical movement representation induced by practice,” J. Neurophysiol, vol. 79, no. 2, pp. 1117–1123, February 1998. [DOI] [PubMed] [Google Scholar]
  • [28].Genthe K, Schenck C, Eicholtz S, Zajac-Cox L, Wolf S, and Kesar TM, “Effects of real-time gait biofeedback on paretic propulsion and gait biomechanics in individuals post-stroke,” Top Stroke Rehabil, vol. 25, no. 3, pp. 186–193, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [29].Choi JT, Vining EPG, Reisman DS, and Bastian AJ, “Walking flexibility after hemispherectomy: split-belt treadmill adaptation and feedback control,” Brain, vol. 132, no. Pt 3, pp. 722–733, March 2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Yanagihara D, Udo M, Kondo I, and Yoshida T, “A new learning paradigm: adaptive changes in interlimb coordination during perturbed locomotion in decerebrate cats,” Neuroscience Research, vol. 18, no. 3, pp. 241–244, December 1993. [DOI] [PubMed] [Google Scholar]
  • [31].Roerdink M and Beek PJ, “Understanding inconsistent step-length asymmetries across hemiplegic stroke patients impairments and compensatory gait,” Neurorehabil Neural Repair, vol. 25, no. 3, pp. 253–258, March 2011. [DOI] [PubMed] [Google Scholar]
  • [32].Malone LA and Bastian AJ, “Thinking about walking: effects of conscious correction versus distraction on locomotor adaptation,” J. Neurophysiol, vol. 103, no. 4, pp. 1954–1962, April 2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Mulroy SJ, Eberly VJ, Gronely JK, Weiss W, and Newsam CJ, “Effect of AFO design on walking after stroke: impact of ankle plantar flexion contracture,” Prosthet Orthot Int, vol. 34, no. 3, pp. 277–92, September 2010. [DOI] [PubMed] [Google Scholar]

RESOURCES