Development of a virtual reality-based intervention for community walking post stroke: an integrated knowledge translation approach.

PURPOSE: To develop a virtual reality (VR) based intervention targeting community walking requirements. METHODS: Two focus groups each involving 7 clinicians allowed exploring optimal features, needed support and perceived favorable/unfavorable factors associated with the use of the VR-based intervention from the clinicians' perspective. Three stroke survivors and 2 clinicians further interacted with the intervention and filled questionnaires related to acceptability and favorable/unfavorable perceptions on the VR intervention. Stroke participants additionally rated their perceived effort (NASA Tax Load Index), presence (Slater-Usoh-Steed) and cybersickness (Simulator Sickness Questionnaire). RESULTS: Results identified optimal features (patient eligibility criteria, task complexity), needed support (training, human assistance), as well as favorable (cognitive stimulation, engagement, representativeness of therapeutic goals) and unfavorable factors (misalignment with a natural walking pattern, client suitability, generalization to real-life) associated with the intervention. Acceptability scores following the interaction with the tool were 28 and 42 (max 56) for clinicians and ranged from 43 to 52 for stroke participants. Stroke participants reported moderate perceptions of effort (range:20-33/max:60), high levels of presence (29-42/42) and minimal cybersickness (0-3/64). CONCLUSION: Findings collected in the early development phase of the VR intervention will allow addressing favorable/unfavorable factors and incorporating desired optimal features, prior to conducting effectiveness and implementation studies.

This study presents the development process of a new virtual reality (VR) intervention for community walking and participation in stroke survivors.Results from the focus group and hands-on pilot trial suggest that the VR intervention is feasible and accepted by clinicians and stroke survivors.Addressing favorable/unfavorable factors and incorporating features desired by clinicians in the development of the VR tool should promote its eventual implementation in clinical setting.

Instantaneous effect of real-time avatar visual feedback on interlimb coordination during walking post-stroke.

BACKGROUND: Gait asymmetry, which is common after stroke, is typically characterized using spatiotemporal parameters of gait that do not consider the aspect of movement coordination. In this manuscript, we examined whether an avatar-based feedback provided as a single-session intervention to improve gait symmetry also improved inter-limb coordination among people with stroke and we examined the relationship between changes in coordination and step length symmetry. METHODS: Twelve stroke participants walked on a self-paced treadmill with and without a self-avatar that replicated their locomotor movements in real time. Continuous relative phase and angular coefficient of correspondence calculated using bilateral sagittal hip movements were used to quantify temporal and spatial interlimb coordination, respectively. Spatial gait symmetry, previously shown to improve with the avatar feedback, was quantified using step length ratio between both limbs, with the largest value as numerator. FINDINGS: Participants who improved their spatial symmetry during avatar exposure also improved their temporal coordination, while spatial coordination remained unchanged. Overall, improvements in spatial symmetry correlated positively with improvements in temporal coordination. The non-paretic hip and paretic ankle angle excursion in the sagittal plane also significantly increased during avatar exposure. INTERPRETATION: Improvements in gait symmetry may be explained by changes in interlimb coordination. Current data and existing literature further suggest that such improvements are largely driven by adaptations in non-paretic leg movements, notably at the hip. By providing real-time information on walking movements not affordable in other ways, avatar-based feedback shows great potential to improve gait symmetry and interlimb coordination post-stroke.

Real-Time Avatar-Based Feedback to Enhance the Symmetry of Spatiotemporal Parameters After Stroke: Instantaneous Effects of Different Avatar Views.

Gait asymmetry, one of the hallmarks of post stroke locomotion, often persists despite gait rehabilitation interventions, impacting negatively on functional mobility. Real-time feedback and biological cues have been studied extensively in recent years, but their applicability to post-stroke gait symmetry remain questionable. This proof-of-concept study examined the feasibility and instantaneous effects of real-time visual feedback provided in the form of an avatar in twelve participants with stroke on gait symmetry and other gait-related outcomes. The visual avatar was presented via three different views from the back, front and paretic side. Avatar feedback from the paretic side view showed significant increase in bilateral step length, paretic swing time and treadmill walking speed, but no significant differences were found in symmetry measures in any of the three views. Those who had changes in symmetry ratio >0 were grouped as responders to spatial symmetry improvement in the side view. The responders had a significantly higher Chedoke-McMaster Stroke Assessment foot score and presented with a larger initial step length on the paretic side. Furthermore, all participants provided positive feedback and no adverse effects were observed during the experiment. Overall, these findings suggest that real-time avatar-based feedback can be used as an intervention to improve post-stroke gait asymmetry.

Healthy young adults implement distinctive avoidance strategies while walking and circumventing virtual human vs. non-human obstacles in a virtual environment.

This study examines how three types of obstacles (cylinder, virtual human and virtual human with footstep sounds) affect circumvention strategies of healthy young adults. Sixteen participants aged 25.2 ±â€¯2.5 years (mean ±â€¯1SD) were tested while walking overground and viewing a virtual room through a helmet mounted display. As participants walked towards a stationary target in the far space, they avoided an obstacle (cylinder or virtual human) approaching either from the right (+40°), left (-40°) or head-on (0°). Obstacle avoidance strategies were characterized using the position and orientation of the head. Repeated mixed model analysis showed smaller minimal distances (p = 0.007) while avoiding virtual humans as compared to cylinders. Footstep sounds added to virtual humans did not modify (p = 0.2) minimal distances compared to when no sound was provided. Onset times of avoidance strategies were similar across conditions (p = 0.06). Results indicate that the nature of the obstacle (human-like vs. non-human object) matters and can modify avoidance strategies. Smaller obstacle clearances in response to virtual humans may reflect the use of a less conservative avoidance strategy, due to a resemblance of obstacles to pedestrians and a recall of strategies used in daily locomotion. The lack of influence of footstep sounds supports the fact that obstacle avoidance primarily relies on visual cues and the principle of 'inverse effectiveness' whereby multisensory neurons' response to multimodal stimuli becomes weaker when the unimodal sensory stimulus (vision) is strong. Present findings should be taken into consideration to optimize the ecological validity of VR-based obstacle avoidance paradigms used in research and rehabilitation.

Ecological Virtual Reality Evaluation of Neglect Symptoms (EVENS): Effects of Virtual Scene Complexity in the Assessment of Poststroke Unilateral Spatial Neglect.

BACKGROUND: Unilateral spatial neglect (USN) is a highly prevalent and disabling poststroke impairment. USN is traditionally assessed with paper-and-pencil tests that lack ecological validity, generalization to real-life situations and are easily compensated for in chronic stages. Virtual reality (VR) can, however, counteract these limitations. OBJECTIVE: We aimed to examine the feasibility of a novel assessment of USN symptoms in a functional shopping activity, the Ecological VR-based Evaluation of Neglect Symptoms (EVENS). METHODS: EVENS is immersive and consists of simple and complex 3-dimensional scenes depicting grocery shopping shelves, where joystick-based object detection and navigation tasks are performed while seated. Effects of virtual scene complexity on navigational and detection abilities in patients with (USN+, n = 12) and without (USN-, n = 15) USN following a right hemisphere stroke and in age-matched healthy controls (HC, n = 9) were determined. RESULTS: Longer detection times, larger mediolateral deviations from ideal paths and longer navigation times were found in USN+ versus USN- and HC groups, particularly in the complex scene. EVENS detected lateralized and nonlateralized USN-related deficits, performance alterations that were dependent or independent of USN severity, and performance alterations in 3 USN- subjects versus HC. CONCLUSION: EVENS' environmental changing complexity, along with the functional tasks of far space detection and navigation can potentially be clinically relevant and warrant further empirical investigation. Findings are discussed in terms of attentional models, lateralized versus nonlateralized deficits in USN, and tasks-specific mechanisms.

Cortical mechanisms underlying sensorimotor enhancement promoted by walking with haptic inputs in a virtual environment.

Sensorimotor integration is a complex process in the central nervous system that produces task-specific motor output based on selective and rapid integration of sensory information from multiple sources. This chapter reviews briefly the role of haptic cues in postural control during tandem stance and locomotion, focusing on sensorimotor enhancement of locomotion post stroke. The use of mixed-reality systems incorporating both haptic cues and virtual reality technology in gait rehabilitation post stroke is discussed. Over the last decade, researchers and clinicians have shown evidence of cerebral reorganization that underlies functional recovery after stroke based on results from neuroimaging techniques such as positron emission tomography and functional magnetic resonance imaging. These imaging modalities are however limited in their capacity to measure cortical changes during extensive body motions in upright stance. Functional near-infrared spectroscopy (fNIRS) on the other hand provides a unique opportunity to measure cortical activity associated with postural control during locomotion. Evidence of cortical changes associated with sensorimotor enhancement induced by haptic touch during locomotion is revealed through fNIRS in a pilot study involving healthy individuals and a case study involving a chronic stroke patient.

Action-perception coupling in kinesthesia: a new approach.

According to recent findings, intentional motor actions are controlled by resetting the referent position, R, at which neuromuscular elements, including reflexes, begin to act. It is suggested that somatosensory afferents inform the brain about the deviation (P) of body segments from the centrally set referent position. To perceive the actual position (Q) of body segments and form the position sense (PS), the central and afferent signals are combined: Q=R+P. In previous studies, the R has been shown to remain invariant during involuntary changes in the wrist position elicited by sudden unloading of muscles, suggesting that only the afferent component is responsible for the PS during this reflex. In contrast, the central PS component, R, is predominantly responsible for PS during intentional motion in isotonic conditions. We tested the hypothesis that the R and P are interchangeable PS components such that involuntary changes in wrist position elicited by the unloading reflex can easily be reproduced by making intentional changes in wrist position in isotonic conditions, in the absence of vision. The PS rule also suggests that PS is independent of sense of effort, which was tested by asking subjects to reproduce elbow joint angles under different constant loads. We also tested the hypothesis that tendon vibration may elicit motion that may not be perceived by subjects (no-motion illusion). These hypotheses were confirmed in three experiments. It is concluded that the R and P are additive components of PS and that, contrary to the conventional view, PS is independent of the sense of effort or efference copy. The PS rule also explains kinesthetic illusions and the phantom limb phenomenon. This study advances the understanding of action-perception coupling in kinesthesia.

Corticospinal control strategies underlying voluntary and involuntary wrist movements.

The difference between voluntary and involuntary motor actions has been recognized since ancient times, but the nature of this difference remains unclear. We compared corticospinal influences at wrist positions established before and after voluntary motion with those established before and after involuntary motion elicited by sudden removal of a load (the unloading reflex). To minimize the effect of motoneuronal excitability on the evaluation of corticospinal influences, motor potentials from transcranial magnetic stimulation of the wrist motor cortex area were evoked during an EMG silent period produced by brief muscle shortening. The motoneuronal excitability was thus equalized at different wrist positions. Results showed that the unloading reflex was generated in the presence of a corticospinal drive, rather than autonomously by the spinal cord. Although the tonic EMG levels were substantially different, the corticospinal influences remained the same at the pre- and post-unloading wrist positions. These influences however changed when subjects voluntarily moved the wrist to another position. Previous studies showed that the corticospinal system sets the referent position (R) at which neuromuscular posture-stabilizing mechanisms begin to act. In self-initiated actions, the corticospinal system shifts the R to relay these mechanisms to a new posture, thus converting them from mechanisms resisting to those assisting motion. This solves the classical posture-movement problem. In contrast, by maintaining the R value constant, the corticospinal system relies on these posture-stabilizing mechanisms to allow involuntary responses to occur after unloading. Thus, central control strategies underlying the two types of motor actions are fundamentally different.

Subthreshold corticospinal control of anticipatory actions in humans.

Previous findings suggest that, by influencing the subthreshold state of motoneurons, the corticospinal pathways can set and reset the threshold position at which wrist muscle recruitment begins. Here we assumed that the corticospinal system can change the threshold position in a similar way before anticipated perturbation to pre-determine an appropriate emerging response to it. We first analyzed motor-evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) applied to the wrist area of motor cortex before unloading of preloaded wrist flexors, i.e. before the subsequent involuntary wrist motion to another position (natural unloading). Subjects then learned to diminish the post-unloading movement extent without activating antagonist (extensor) muscles before unloading or making intentional movement corrections after unloading (adjusted unloading). Although activity levels of wrist muscles before unloading were similar, MEPs of extensor but not pre-loaded flexor muscles were higher before adjusted unloading. We also applied TMS in combination with a torque pulse that shortened extensor muscles such that the MEP occurred when the motoneuronal excitability was minimized. Although diminished following muscle shortening, MEPs before adjusted unloading were still higher than before natural unloading. Results suggest that the corticospinal system, possibly together with other descending systems participated in the tonic subthreshold facilitation of antagonist motoneurons before adjusted unloading, which appears sufficient in modifying motor commands and motion leading to adjusted unloading. This study reinforces previous findings that descending systems, in particular, the corticospinal system can employ threshold position control during and after learning a novel action.

Multijoint reflex responses to constant-velocity volitional movements of the stroke elbow.

Multijoint reflex coupling could impact the voluntary control of functional arm movements in people post stroke. The multijoint responses to active-assist, constant-velocity movements of the elbow joint were measured in 14 individuals post stroke and 9 neurologically intact controls. Resulting responses in the stroke group illustrated a change in the reflex coupling of the elbow and shoulder muscles compared with passive perturbations of the spastic elbow. Voluntary effort during constant-velocity elbow extension resulted in reflex shoulder abduction, differing from the reflex coupling observed between the elbow flexors and shoulder adductors observed during passive elbow extension. These results suggest that post stroke, voluntary drive alters reflex coupling of the elbow and shoulder. Flexion of the elbow during active-assist also resulted in reflex coupling. Shoulder abduction torque decreased with constant-velocity flexion of the elbow; however, no net adduction was observed at the end of the perturbation. Shoulder flexion/extension and internal/external rotation torque responses demonstrated similar modulations to imposed active-assist perturbations of the elbow in subjects post stroke. Responses were absent during passive perturbations of the control elbow; however, shoulder torque modulations were observed during constant-velocity, active-assist tasks. The active-assist response patterns in controls were similar to stroke subjects during the extension task but opposite during flexion of the elbow. This study provides evidence of a neural coupling between elbow and shoulder muscles and a modulation of this coupling during voluntary drive of the spastic arm.

Multijoint reflexes of the stroke arm: neural coupling of the elbow and shoulder.

The reflex torque responses of the elbow and shoulder to constant velocity angular extensions of the full comfortable range of the spastic elbow were measured in 16 people with unilateral stroke and 6 neurologically intact controls in order to identify the interjoint reflex coupling that occurs after stroke. The resulting responses showed a substantial reflex torque at the elbow and shoulder in subjects with stroke, with 12 of the 16 subjects producing adduction of the shoulder in response to passive extension of the elbow. The presence of simultaneous shoulder flexion torque with elbow flexion torque and with an identical waveform indicated an active role of biarticular elbow/shoulder flexors, such as the biceps. As the biceps muscle produces a shoulder abduction moment, shoulder adduction produced during elbow extension was thought to be associated with neural rather than biomechanical coupling. These results suggest that spasticity in people with stroke is more complex than its traditional perception as a hyperexcitable stretch reflex, and includes potent heteronymous reflex pathways. The reflex coupling observed between the shoulder and elbow should be considered in the diagnosis and clinical management of spasticity. The potential impact of this reflex on the coordination of volitional arm movements will be examined in future studies.

Reliability of biomechanical spasticity measurements at the elbow of people poststroke.

OBJECTIVE: To determine the minimum number of measurements required to obtain a reliable estimate of upper-extremity spasticity using biomechanic assessment across multiple testing trials and dates. DESIGN: Single-center, longitudinal study with repeated measurements of spastic upper-extremity torque measures taken 1 week apart. SETTING: A hospital-based laboratory with an isokinetic testing system. PARTICIPANTS: Sixteen subjects more than 6 months poststroke with upper-extremity spasticity. INTERVENTION: Elbow flexor hypertonia was assessed with a custom-made manipulandum attached to a 6-axis load cell and a Biodex System 3 isokinetic testing machine. Movements into extension were imposed at 4 speeds: 6 degrees /s, 30 degrees /s, 60 degrees /s, and 90 degrees /s. MAIN OUTCOME MEASURES: The resistive torque and electromyographic response to these imposed movements were measured. The torque response at the slowest speed (6 degrees /s) was attributed solely to the passive elements of the elbow and was subtracted from the torque response at the higher speeds (30 degrees /s, 60 degrees /s, 90 degrees /s), leaving only reflex torque. The reflex torques at 30 degrees /s, 60 degrees /s, and 90 degrees /s were used for further analysis. Peak torque, peak joint stiffness, and onset angle of reflex torque responses were found; repeatability and daily variability of these measures were statistically examined. The variabilities due to the subject, test day, and trial number were computed. The overall reliability of each parameter at the 3 higher test speeds using different testing schemes was also calculated. RESULTS: Ninety percent reliability in the measurement of all parameters was obtained after at least 2 days of testing during which 3 tests a day were performed. The variability in between-subjects measurements was at least 4 times greater than the intertrial variability when testing at the highest speeds; daily variability that was up to 50% of the intersubject variability was also observed. The biomechanic measures correlated well with the Ashworth Scale (Spearman rho=.84, P<.005), a clinical measure of hypertonia. CONCLUSIONS: We recommend at least 2 test dates to account for the daily variability in the spastic reflex response and to ensure reliable spasticity measurements.