Electrotactile Feedback of Sway Position Improves Postural Performance during Galvanic Vestibular Stimulation

Transcription

1 BASIC AND CLINICAL ASPECTS OF VERTIGO AND DIZZINESS Electrotactile Feedback of Sway Position Improves Postural Performance during Galvanic Vestibular Stimulation Scott J. Wood, a F. Owen Black, b Hamish G. MacDougall, c and Steven T. Moore d a Universities Space Research Association, NASA Johnson Space Center, Houston, Texas 775, USA b Neurotology Research, Legacy Health System, Portland, Oregon 97, USA c Department of Psychology, University of Sydney, Sydney, Australia d Department of Neurology, Mount Sinai School of Medicine, New York, New York 9, USA The purpose of this study was to assess the influence of electrotactile feedback on postural control performance during binaural galvanic vestibular stimulation (GVS). Postural equilibrium was measured with a computerized hydraulic platform in healthy adults (6M, F, 65 y). Feedback of anterior posterior (AP) and mediallateral (ML) body sway was derived from a -axis linear accelerometer mounted on a torso belt and displayed on a -point electrotactile array held against the anterior dorsal tongue. Subjects were trained to use the tongue electrotactile feedback () by voluntarily swaying to draw figures on their tongue, both with and without GVS. Subjects performed randomized trials (-s duration with eyes closed, trials per condition), including support surface conditions (fixed, rotational sway-referenced, translating the support surface proportional to AP sway, and combined rotational translational support-platform sway referencing), and feedback conditions (baseline, GVS, and GVS with ). Postural performance was assessed using deviations from upright (peak-to-peak and root-mean-square sway) and convergence toward stability limits (time and distance to limit of support boundaries). Postural stability was impaired (with respect to baseline) during GVS in all platform conditions, with larger decrements in performance during trials with rotation sway-referencing. Electrotactile feedback improved performance with GVS toward non-gvs levels, especially during trials with rotation sway-referencing. These results demonstrate the effectiveness of in providing sensory substitution to maintain postural stability during vestibular disturbances. Key words: posturography; translation; biofeedback Introduction Integration of multisensory inputs to detect tilts relative to gravity is critical for sensorimotor control of upright orientation. Bilat- Address for correspondence: F. Owen Black, Neurotology Research, Legacy Health System, 5 NE nd Ave., Portland, OR 97. fob@neurotology.org eral vestibular loss leads to difficulty in reliably making judgments of perceived verticality, and adversely affects the stabilization of head and body posture., MacDougall and colleagues recently demonstrated that unpredictably varying binaural galvanic vestibular stimulation (GVS) induces postural instability that is similar to profound bilateral loss, especially during conditions where visual and proprioceptive feedback are altered. Hlavacka and Basic and Clinical Aspects of Vertigo and Dizziness: Ann. N.Y. Acad. Sci. 6: 9 9 (9). doi:./j x C 9 New York Academy of Sciences. 9

2 Wood et al.: Electrotactile Feedback Improves Posture during GVS 9 Horak 5 also demonstrated that binaural GVS compromises postural stability during translations of the support surface. One advantage of GVS is the transient nature of the impairment. MacDougall, Moore, and colleagues have observed that short exposures to GVS can temporarily cause postural and locomotor 6 dysfunction without any evidence of aftereffects when the GVS is turned off. Thus, GVS can be employed as a technique to briefly disrupt vestibular-mediated responses while allowing comparison to baseline measures,and/orcanbeusedtoevaluatepromising treatments such as sensory substitution aids. Rehabilitating patients with bilateral vestibular loss typically involves training to utilize vision, proprioception, and peripheral sensation to substitute for the missing vestibular input. 7, However, compensation is limited in some patients. 9 Retraining proprioceptive function is often incomplete due to the strong reliance on visual mechanisms to compensate for vestibular loss. Electrotactile feedback to the tongue was developed by Bach-y-Rita and colleagues as a sensory aid to display orientation cues on the heavily innervated and highly sensitive dorsal anterior tongue surface. Initially devised to examine sensory substitution in the blind, this technique has recently been extended to provide a substitute body-orientation reference to subjects with vestibular dysfunction. The purpose of this study was to assess the influence of tongue electrotactile feedback () on postural control performance during binaural GVS as previously developed by MacDougall and colleagues. Trials were conducted with eyes closed and altered somatosensory feedback to accentuate reliance on the electrotactile feedback in the presence of the disrupted vestibular function associated with GVS. Methods Postural performance was measured with a computerized hydraulic platform in healthy adults (6M, F, 65 y). Each subject reported no history of balance or vestibular abnormalities on a medical history questionnaire, and performed within normal limits on a standardized posture test. All experimental procedures were approved by the Legacy Institutional Review Board, and all subjects provided informed consent prior to inclusion. Subjects were instructed to maintain upright stance with arms folded and eyes closed for -s trials. Subjects stood on a movable platform with shoes off, feet shoulder width apart, and ankle joints aligned with the rotational axis of the support surface. Straps from an overhead safety harness were configured so that they did not bear any weight, and sway was unimpeded within normal limits of stability. Displacements of the support surface were measured with potentiometers coupled to the motors that drove rotational or linear movement of the platform. Center-of-mass (COM) sway angles were derived from anterior posterior (AP) and medial lateral (ML) center-of-force positions using a low-pass Butterworth filter (second order, cutoff frequency at.5 Hz), with the height of the COM estimated at 55% of the subject height. Rotational sway-referencing involved rotating the support surface by the same angle as the instantaneous AP sway so that ankle proprioceptive feedback was altered. In a similar fashion, translational sway-referencing involved moving the support surface forward or backward in direct proportion to AP sway (. cm/deg). For example, leaning forward would result in forward acceleration of the support surface. Four support-surface conditions were utilized: fixed, rotational sway-referenced (SR), translational SR, and combined rotational translational SR. Three feedback conditions included: baseline without GVS, GVS without, and GVS with. These conditions ( support feedback) were block randomized across subjects so that there was an alternating pattern of baseline, GVS, and conditions, which was then repeated again in the same order for a total of trials. Trials in which subjects raised their arms or moved their feet were marked as falls and repeated.

3 9 Annals of the New York Academy of Sciences Galvanic Vestibular Stimulation GVS was achieved using a system previously described. Bilateral bipolar current (max.5 ma) was delivered to the surface of the subject s skin via leads and large electrodes placed over the mastoid processes. An elastic headband was utilized to maintain consistent surface contact of the electrodes throughout the session. The large size of the electrodes and use of electrode gel ensured that the stimulus amplitude did not cause discomfort, and the lead cables were arranged so they did not restrict movements. The galvanic stimulus was a modified sum-of-sines that was delivered in bilateral and bipolar fashion. The dominant frequencies were at.6,.,., and.6 Hz. The GVS stimulus was initiated just prior to the start of a trial by a stand-alone laptop computer, and turned off at the end of each trial (typically 5- to -s periods of constant exposure). Motion-sickness symptoms, if any were noted, were recorded in between trials. Tongue Electrotactile Feedback Electrotactile feedback of AP and ML body orientation was derived from a two-axis linear accelerometer mounted on a torso belt. The linear accelerometer package was mounted at approximately 55% of subject height to be near the COM, thus the feedback closely corresponded to body-sway position that was used for the rotational and translation swayreferencing. The display consisted of a - point electrotactile array held against the anterior dorsal tongue (BrainPort, Wicab, Inc., Middleton, Wisconsin, USA). The signals were scaled so that instantaneous feedback of sway was continuously provided within the limits of stability. The intensity of the electrotactile feedback was adjusted as needed throughout the session to optimize resolution while minimizing discomfort. Subjects were trained to use at the beginning of each session by voluntarily swaying to draw figures on their tongue, including a cross, circle, figure, and letters. The training trials were then repeated using both and GVS. In order to minimize stabilizing influences of cables connected to the control unit, the cables from the tongue array were secured to the torso belt and thus moved with the subject. Subjects were required to keep the intraoral display in their mouths on all trials, including those where was not used. Data Analysis Postural performance was assessed using deviations from upright (peak-to-peak and rootmean-square [RMS] sway) and convergence toward stability limits (minimum time and distance to limit of support boundaries). The limit of support (LOS) in the AP plane was defined by the foot boundary. 5 AP sway was differentiated to compute sway velocity, and both minimum time and distance to LOS took into account the direction of COM, for example, the distance or time to the LOS boundary was computed relative to either edge of toes or heel, depending on which the COM was moving toward. A maximum value of 5 s for the time to LOS was allowed. Using an alpha error of.5 as the decision rule, the null hypothesis that there was no difference across conditions was tested for each of these measures using a paired Wilcoxon signed-rank statistic. Results The GVS was well tolerated by all subjects. Two subjects reported mild flushing and sweating, and one subject noted mild stomach awareness. During quiet stance without swayreferencing, most subjects perceived more unsteadiness during GVS laterally than in the AP direction. However, the most striking differences occurred in the direction of the swayreferencing, so all remaining results refer to AP measures. No falls were observed during baseline testing of each condition, nor during GVS testing

4 Wood et al.: Electrotactile Feedback Improves Posture during GVS 95 A Fixed C Rotation ** * * Baseline GVS Baseline GVS B Translation D Rotation-Translation * Baseline GVS Baseline GVS Figure. Comparison of anterior posterior (AP) peak-to-peak sway for each of the four support and three feedback conditions (mean ± SEM). Note that indicates significant difference between conditions at P <.5 level ( indicates P <.). with a fixed or translation-sr support surface. However, two falls occurred during GVS testing for each condition where there was rotation- SR support surface. The peak-to-peak sway was generally twice as large during conditions involving rotation-sr as compared to conditions that did not involve rotations (compare Fig. C and D with Fig. A and B). Consistent with the results of MacDougall and colleagues, GVS significantly increased peak-to-peak sway relative to baseline for both fixed and rotation-sr conditions (Fig. A and C). The same finding was present for RMS sway (e.g., increase from. ±. to. ±. for rotation-sr, mean ± SEM). GVS also significantly decreased the minimum time to LOS for both conditions (Fig. A and C), indicating that during trials with GVS, subjects came closer to stability margins. This was also true for the minimum distance to LOS, for example, decreasing from. ±. to.9 ±. for rotation-sr. The same trend for GVS to impair postural performance was present during the conditions involving translation-sr, although this was only significant for peak-to-peak and RMS sway during rotation translation SR (Fig. D). Most notable to this investigation, electrotactile feedback during GVS successfully improved peak-to-peak and RMS sway to baseline levels for all conditions (Fig. ). There was a significant improvement between GVS without and GVS with for both rotation-sr and translation-sr conditions. Interestingly, did not significantly alter measures of minimum time or distance to LOS during GVS. Discussion Thefirstmainfindinginthisstudywasthat postural stability was impaired with GVS in all platform conditions, with larger decrements in performance during trials with rotation swayreferencing. This is consistent with the findings of MacDougall and colleagues, who utilized a larger stimulus current (5 ma versus.5 ma). In addition, this study also demonstrates that

5 96 Annals of the New York Academy of Sciences A Fixed * * C Rotation * * Baseline GVS Baseline GVS B Translation D Rotation-Translation * Baseline GVS Baseline Figure. Comparison of minimum time to limit of support (LOS) for each of the four support and three feedback conditions (same notations as in Fig. ). GVS GVS compromises postural stability as determined by convergence to stability limits. The minimum time to LOS reflects control of both COM position and velocity, 5 and therefore may better account for fall risk during dynamic motions. 6 The observation that subjects were more at risk of exceeding their stability limits is consistent with MacDougall s conclusion that higher levels of GVS distort vestibular input to balance centers in the cerebellum. The second main finding of this study was that electrotactile feedback improved performance during GVS toward baseline levels, again with the greatest improvement during trials with rotation sway-referencing. This is consistent with Vuillerme and colleagues, who recently demonstrated that tongue electrotactile feedback reduced sway in normals on both fixed and unstable (foam block) support surfaces. 7 Our results extend the work of Vuillerme in that we demonstrate electrotactile feedback is also effective during transient vestibular impairment such as that induced by GVS.,6 Electrotactile feedback has also been effective in stabilizing balance in chronic bilateral vestibular loss (BVL) patients., Our laboratory has also noted improvements using with BVL patients (unpublished data). Anecdotally, training appeared to have retention effects that appear promising. Figure provides preliminary data for one of our BVL patients during the eyes-closed rotation-sr condition, commonly referred to as Sensory Organization Test 5. Prior to training, this subject exhibited a fall pattern that is typical of bilateral loss of vestibular function. The subject was then exposed to four consecutive days of training with both morning and afternoon sessions. Following the training, retention effects were monitored with measurements immediately after the last training, and then at + h and +,,, and 5 days. After the first day of training, although improvements were noted, the subject still tended to fall on SOT-5. However, by the end of the second day the median SOT-5 score was already within clinical normal limits, with further improvements after

6 Wood et al.: Electrotactile Feedback Improves Posture during GVS 97 Median EQ Score Fall Days Training Post-Training Figure. Sample of training and retention effects of tongue electrotactile feedback () on one bilateral vestibular loss (BVL) patient during the rotation sway-referenced (SR) condition. Each bar represents the equilibrium (EQ) score (median of three trials) derived from peak-to-peak anterior posterior (AP) sway. An EQ of represents a fall, while EQ of represents absence of sway. The line at 5 represents the age-adjusted clinical normal limits. a third day of testing. Performance appeared to plateau by the fourth day, and remained within clinical normal limits throughout the posttraining period (through at least 5 days). It is interesting to note that other tactile feedback methods have shown similar retention effects. Given the high correlation between postural performance in normals during GVS and postural performance in BVL patients, the ability to reduce sway in both groups using electrotactile feedback further confirms the utility of GVS as a model for functional impairment of vestibular origin. However, the use of sensory substitution for balance control depends on the environmental context and individual sensory weighting, 5,9 as well as on the nature of the feedback provided. It is interesting to note that did not significantly improve stability measures that account for COM position and velocity control toward stability margins, such as minimum time to LOS. This may be explained in that our feedback was based strictly on position deviations from upright. One extension of our study would be to alter the electrotactile feedback to predict sway toward stability margins based on both sway position and velocity, and then evaluate the influence of this feedback on minimum time to LOS during GVS. Our study introduced a novel method of sway-referencing involving translations of the support surface in response to sway from upright. Hlavacka and Horak 5 observed that bilateral GVS resulted in greater forward displacements during backward platform translations. The increase in these displacements relative to a given increase in GVS current (.5- to -ma range) was interpreted as vestibulospinal sensitivity, which correlated to the severity of somatosensory loss in the feet due to diabetic neuropathy. The polarity of our translations (e.g., backward acceleration during backward lean) would have the tendency to counteract body sway. This explains why the sway measures during translation SR were similar to the fixed support condition. We propose that changing the gain of the translation sway-referencing, or even the polarity of the displacements, will provide additional insight into the use of somatosensory feedback for posture control. In summary, these results demonstrate the effectiveness of in providing sensory substitution to maintain postural stability with distorted vestibular input. The improvements in postural performance reflect the support surface conditions and the specific information displayed to the subject. GVS provides a useful mechanism to optimize and validate sensory substitution feedback strategies targeted toward rehabilitating vestibular-mediated function. Acknowledgments This work was supported by the National Space Biomedical Research Institute through NASA NCC 9-5 (NA 7), and is dedicated to the memory of Paul Bach-y-Rita. The authors gratefully acknowledge the use of a prototype BrainPort balance device from Wicab, Inc. (Middleton, Wisconsin, USA); and Valerie Stallings for data-collection support.

7 9 Annals of the New York Academy of Sciences Conflicts of Interest The authors declare no conflicts of interest. References. Bronstein, A.M. et al Visually and posturally mediated tilt illusion in Parkinson s disease and in labyrinthine defective subjects. Neurology 7: Gresty, M.A. et al. 99. Neurology of otolith function. Peripheral and central disorders. Brain 5: Black, F.O. et al. 9. Abnormal postural control associated with peripheral vestibular disorders. Prog. Brain Res. 76: MacDougall, H.G. et al. 6. Modeling postural instability with galvanic vestibular stimulation. Exp. Brain Res. 7:. 5. Hlavacka, F. & F.B. Horak. 6. Somatosensory influence on postural response to galvanic vestibular stimulation. Physiol. Res. 55(Suppl. ): S S7. 6. Moore, S.T. et al. 6. Modeling locomotor dysfunction following spaceflight with galvanic vestibular stimulation. Exp. Brain Res. 7: Brandt, T Bilateral vestibulopathy revisited. Eur. J. Med. Res. : Foster, C.A. 99. Vestibular rehabilitation. Bailliere s Clin. Neurol. : Black, F.O., S.W. Wade & L.M. Nashner What is the minimal vestibular function required for compensation? Am. J. Otol. 7: 9.. Bach-Y-Rita, P.. Tactile sensory substitution studies. Ann. N.Y. Acad. Sci. : 9.. Tyler, M., Y. Danilov & P. Bach-Y-Rita.. Closing an open-loop control system: vestibular substitution through the tongue. J. Integr. Neurosci. : Black, F.O.. What can posturography tell us about vestibular function? Ann. N.Y. Acad. Sci. 9: Peterka, R.J. & F.O. Black. 99. Age-related changes in human posture control: sensory organization tests. J. Vestib. Res. : Winter, D.A.. Biomechanics and Motor Control of Human Movement. John Wiley & Sons. New York. 5. Pai, Y.C. & J. Patton Center of mass velocityposition predictions for balance control. J. Biomech. : Forth, K.E., E.J. Metter & W.H. Paloski. 7. Age associated differences in postural equilibrium control: a comparison between EQscore and minimum time to contact (TTC(min)). Gait Posture 5: Vuillerme, N. et al.. Sensory supplementation system based on electrotactile tongue biofeedback of head position for balance control. Neurosci. Lett. : 6.. Wall, C., rd. et al.. Applications of vibrotactile display of body tilt for rehabilitation. Conf. Proc. IEEE Eng. Med. Biol. Soc. 7: Dozza, M., F.B. Horak & L. Chiari. 7. Auditory biofeedback substitutes for loss of sensory information in maintaining stance. Exp. Brain Res. 7: 7.