Moerchen, Victoria A. PhD, PT; Hoefakker, Heather L. DPT
Treadmill training, while not yet standard physical therapy practice with infant populations, is an available intervention with a strong theoretical and empirical base.1,2 Notably, infant treadmill stepping supports task-specific, repetitive leg motion, and provides a context for a high dose of physical activity in short bursts of time, achievable well in advance of the ability to walk. The application of the treadmill to accelerate the acquisition of gait has been studied as a parent-delivered home program conducted separately from and in addition to ongoing, traditional early intervention physical therapy for toddlers with Down syndrome.1 The use of the treadmill to support increased leg activity with or without the expectation of accelerated acquisition of gait has also been reported. Recently, Ulrich and colleagues3,4 have begun to examine the utility of the treadmill to promote increased leg activity in infants with spina bifida (SB). The decrease in leg activity after birth within this population5 supports a need for an intervention such as the treadmill that, at a minimum, fosters increased leg activity at very early ages. The task-specific sensory input that the treadmill provides may also support the activity-dependent plasticity of the spinal cords of these young infants.
The incomplete nature of the spinal lesion in SB interrupts and reduces, but in most cases does not eliminate, supraspinal influences. Peripheral afferent input to the spinal cord is, therefore, critical not only for volitional control below the level of the spinal lesion6 but also to provide input to the residual supraspinal pathways. The role of peripheral afferent input has not been studied as extensively in infants with SB as in adults with traumatic spinal cord injury (SCI), and yet the plasticity of the spinal cord in the developing infant would suggest that such studies are highly relevant to this population. Edgerton et al7 identified that for adults with traumatic SCI, stepping on a treadmill alone is not sufficient for neuroplastic recovery, but that spatial-temporal coupling of sensory input with stepping was essential to foster change in locomotor control. Sensory enhancements within the treadmill context have been shown to augment the step responsiveness of infants with typical development,8 Down syndrome,9 and SB.4 The mechanism of effect for this augmentation during stepping is thought to extend from repetitive cycles of perception-action coupling that allow for augmentation of confluent input through visual, tactile, and proprioceptive receptors. Spinal interneurons are capable of integrating sensory information with residual descending pathways to support stepping in individuals with SCI for whom voluntary leg activity would not predict an ability to step.10,11 Similarly, infants with SB showed an immediate increase in step responsiveness to treadmill enhancements that increased hip extension and provided optic flow during stepping, supporting the premise that spinal and descending pathways can be influenced in this population.4
The purpose of this study was to describe the immediate motor responsiveness of infants with SB to single, combined, and manual assistance sensory augmentation, to expand the single sensory augmentation previously described, and to identify important sensory conditions for further study of treadmill stepping with this population. We report on a subset of infants with SB from a multisite, cross-sectional study of single sensory enhancement during treadmill stepping,4 for whom additional data were collected under conditions of combined sensory enhancements and manual assistance but not previously reported. The within-subjects design of this study allow us to consider the responses of the same infants to 3 sets of sensory augmentation.
From a sample of 12 infants with SB for whom immediate step responsiveness was tested across 3 sets of sensory conditions, data were complete and usable for a subset of 6 infants. Criteria included at least 5 steps taken at each laboratory visit, crying or eye closure of less than 20 seconds during all visual flow trials, and 2 digital videos per trial for all 3 visits. All infants were recruited through an SB specialty clinic and were from white families. Half of the infants were first born and half were second or third in birth order within their families. Table 1 provides the clinical and developmental characteristics of these infants. This study was approved and monitored through the institutional review boards at the University of Wisconsin–Milwaukee and the Children's Hospital of Wisconsin.
Each infant came to the laboratory once per week for 3 sequential weeks. At the first visit, we collected information about the infant's medical history, including lesion level, surgeries, and all associated conditions and secondary complications. At the following 2 visits, medical and developmental information was updated, as relevant for the infant.
Preparation of the infants at each laboratory visit was the same. We removed the infant's clothing down to the diaper and placed reflective markers, 8 mm in diameter, on the iliac crest, greater trochanter, lateral epicondyle, lateral malleolus, and ventral surface of the second metatarsophalangeal joint of both lower extremities to provide data about joint/limb segment position. To examine muscle activity during stepping, we positioned bipolar surface electrodes over the muscle bellies of the quadriceps femoris, tibialis anterior, lateral hamstring, and gastrocnemius/soleus muscles of both lower extremities. The ground electrode was placed over the sacrum. Black cotton tights with the feet cut off and with holes cut around markers were applied over the infants' lower limbs to reduce preoccupation with the electrodes.
The testing procedure at each weekly laboratory visit was the same, with only the set of sensory enhancements varying per visit. Treadmill testing included two 30-second trials per sensory condition, with 5 conditions tested per visit, as detailed later. After treadmill trials were completed, each infant's motor development was assessed using the Bayley Scales of Infant Development III (Table 1).12
The setup for the treadmill trials, including motion capture cameras and digital cameras, was the same across all visits. A custom-designed infant treadmill (Carlin's Creations; 67.5-cm long × 44-cm wide × 29-cm high) was positioned on the top of a table (76.2-cm high) to optimize data capture. For all trials, the examiner held each infant in supported standing with feet touching the treadmill belt surface to allow each infant to take weight as possible (Figure 1). Manual support provided in this manner maximized each infant's opportunity for sensory information via contact of the foot with the treadmill belt and was responsive to the motor abilities and variation in infant behavior within and across trials.2 Six motion capture cameras (Eagle camera; Motion Capture Corp, Santa Rosa, California) were wall-mounted and positioned around the capture space, and the examiner stood holding the infant at either 0° or at 180° within this area, depending on trial condition. For all trials without optic flow enhancement or unloading, the infant was held at 0°, facing the examiner, and 2 digital video cameras were positioned at 90° and 270°, for right and left sagittal plane recording of stepping behaviors. The position of the infant, facing the examiner, was consistent with previous literature on the use of the treadmill with infant and toddler populations.1–4 For trials with optic flow enhancement or unloading, the infant was held at 180°, facing away from the examiner to enhance the infant's view of the treadmill belt surface or to position the infant at the edge of the treadmill belt for rapid unloading of the infant's legs. For optic flow trials, the position of the digital cameras was slightly different, with 1 digital camera positioned on the infant's side (270°) and the other at 0°, to record the infant's visual attention as well as stepping behavior. The position of the infant (held from behind by the examiner) was determined to be optimal for visual attention in the direction of the treadmill belt without producing a reduction in infant stepping behavior.8 Regardless of the infant-to-examiner position, however, the infant's parent was always within visual and auditory range of the infant to ensure infant comfort and to support interaction to keep the infant motivated for stepping. Kinematic, analog, and real-time digital video data were captured using a capture rate of 60 Hz. Only behavioral data from real-time digital video are reported here, as the kinematic and electromyographic data were not usable for all infants.
Three sets of sensory enhancements were examined in this study: single, combined, and manual assistance. The single sensory conditions have previously been described in detail4 and are described here because they form the basis for the combined sensory conditions. The data reported in this study do not include the Velcro trials reported by Pantall et al,4 because Velcro was not used in the combined sensory conditions conducted at this site. The combined sensory conditions were selected after piloting all possible pairings of the single sensory conditions reported by Pantall et al.4 Three 6-month-old infants, 1 developing typically and 2 with SB (L3-4), participated in this pilot work to determine the most salient combinations of enhancements that were then used in this study. Finally, the manual assistance conditions were chosen to examine the response of infants with SB to the locomotor guidance approaches used to retrain stepping in studies with spinalized animals13,14 and in rehabilitation for adults with SCI.15 Specifically, manual assistance was provided at an “optimal” treadmill belt speed (1.6 Hz) as in the single and combined sensory trials, and at a “faster” speed (2.2 Hz), where both speeds were derived from previous work with infants with SB.3
Figure 2 provides an overview of the study design. The order of exposure to sensory sets (single, combined, manual assistance) was randomly assigned per infant (1 sensory set per week for 3 weekly visits), and the sensory conditions within each set were randomly ordered within each visit. The conditions within sensory sets are described in Table 2.
Data reduction included behavior coding of the digital video files (2 videos per trial per infant), using frame-by-frame analysis across 1800 frames per trial. Four coders who were blind to the purposes of the study were used across the study, and each coder had to achieve at least 90% agreement through comparison of their work with that of previously validated coders for the same set of 6 training trials.
Step types and step events were coded from the digital video files, and the frequency of steps was normalized to the total number of steps each infant took across all trials during each laboratory visit (ie, within a set of sensory conditions). Step coding included differentiating steps into 4 types2: single (a step produced by 1 leg does not temporally overlap with a step produced by the other leg), alternating (a step produced by 1 leg does temporally overlap with a step produced by the other leg), parallel (tight temporal overlap between legs, almost like a hop even if asymmetrical), and double (an extra step by 1 leg in an otherwise-alternating sequence). Only total and alternating steps were used for analyses. Total steps reported in this study include a summation of all step types listed previously. Alternating steps are used as a measure of increased complexity of stepping.
Overall activity was coded for all trials to examine active body movement on the treadmill that may or may not have included stepping (head, trunk, or limbs). This variable reflected the clinical value of increased physical activity for a population of infants who are described as less active than their typically developing peers.5,16 Overall activity was coded every 300 frames (5 seconds) and was scored dichotomously using a scale of 0 and 1, where 0 represented no movement and 1 represented clear movement. Scores were combined from both trials of a given sensory condition, and these scores were then converted to percentages per condition.
We asked 2 primary questions: First, does combining sensory inputs increase the immediate step responsiveness in infants with SB? Second, does manual assistance increase the immediate step responsiveness in infants with SB? Within each broad question, we examined total steps and alternating steps. Given the clinical population and small sample size, statistical analyses were accomplished using Wilcoxon matched pair rank tests, and an α level of .05 was defined as statistically significant. SPSS 19 statistical software was used for statistical analyses.
Combined Sensory Enhancements
We conducted an initial examination of the group data, comparing related combinations of single sensory and combined sensory conditions with each other and with the nonenhanced condition (Figure 3). Friction (F) and friction + load (F+L) were observably more salient than the other sensory enhancements or than the nonenhanced condition for both total and alternating steps. Our first question was whether combining sensory inputs would increase the immediate step responsiveness of infants with SB. A Wilcoxon signed rank test was conducted to evaluate whether step responsiveness was indeed greater in the F+L condition than in the nonenhanced condition. The results indicated a significant difference, Z = −2.201, P = .028, for both total steps and alternating steps. Indeed the observed increase in total steps (Figure 3A) was a function of the increase in alternating steps (Figure 3B). The frequency of steps taken in the F condition alone, although clearly salient, did not reach significance compared to the nonenhanced condition (Z = −1.363, P = .173). No other conditions warranted statistical examination.
When we examined the individual profiles for each infant (Figure 4), total steps taken in the F+L trials exceeded the nonenhanced trials in 5 of the 6 infants, which included those with both lumbar and sacral level lesions. Although variability predominated and the sample size was small, it is notable that the greatest responsiveness to the enhancements was observed in 2 of the 3 infants with lumbar level lesions, compared with the group with sacral level lesions.
Because several of the sensory enhancements used in this study included methods that might have reduced the completeness of the step motions in these infants by potentially challenging the ease with which a limb could be fully advanced (eg, load, friction), overall activity was coded per condition as an additional measure of infant responsiveness to the augmented sensory input (Figure 5). We examined overall activity rather than leg activity alone because identifying contexts to increase the physical activity of these infants is highly relevant to this clinical population, beyond just addressing task-specific (potentially gait-relevant) activity such as stepping. While infants were on average more active in the F+L trials, the increase in activity over the nonenhanced trials was minimal and did not reach statistical significance.
The robustness of the treadmill context is arguably in the opportunity it provides for infants to search for a solution to stepping. Although the behavior coding scheme used includes 4-step types, the transition to more complex, alternating stepping is of primary interest for outcomes related to overground gait acquisition.2 With this set of conditions, we hypothesized that manual assistance with stepping would provide sensory input appropriate to the pattern of the rhythmic alternation of stepping and would support the infants' subsequent selection of a more consistent and complex pattern of stepping than they might readily identify and select without manual assistance. Our data supported this hypothesis for the slower of the speeds at which manual assistance was provided (Figure 6). A Wilcoxon test showed that the frequency of alternating stepping increased significantly with manual assistance provided at 48 beats per minute and with the treadmill belt moving at a comfortable speed for infants (0.16 m/s), Z = −2.023, P = .043. The difference in total steps did not reach significance (Z = −0.944, P = .345), nor did the step frequencies after manual assistance at the faster speed (Z = −1.48, P = .138).
The purpose of this study was to describe the immediate motor responsiveness of infants with SB to single, combined, and manually assisted sensory enhancement during treadmill stepping and to identify important sensory conditions for further study of treadmill stepping with this population. Despite the small sample size, the strength of this study was the within-subjects design across 3 sets of sensory enhanced conditions. Our findings supported the robustness of the friction condition observed in the larger data set reported by Pantall et al4 and extended the findings of that study with the identification of F+L as a salient combination of sensory inputs. This study additionally provides preliminary evidence for an immediate increase in the frequency and complexity of stepping in infants with SB following 30 seconds of manually assisted stepping.
Although we examined immediate step responses in infants, the sensory conditions examined in this study were consistent with several of the locomotor training principles described by Behrman and Harkema15: (1) the speed of treadmill belt translation used in all trials (except for one manual assistance trial where treadmill belt translation was at a higher speed) was previously determined to be optimal for infants in these age ranges3 and meets the principle of stepping velocities at normal walking speeds in an infant population17; (2) the load condition, whether alone or as a combined sensory input, increased the load available during stance, and the manually assisted trials included foot placement that augmented stance in a way the infants may not have achieved on their own, given variable foot placements4; (3) the infants' bodies were held upright; although different from adult gait retraining, the infants were allowed to explore stepping while flexing or extending their trunks, going limp, arching, or looking around, allowing them to explore the degrees of freedom available to them in this task2; and (4) the approximation of normal limb kinematics and synchronization of alternating stepping were left to be selected by the infants under nonenhanced treadmill stepping conditions but were aided through the use of the friction belt and were integral to the manually assisted trials. Within this framework, this study took 2 broad sensory approaches: the first included single and combined sensory inputs that provided sensory augmentation as the infant interacted with the treadmill environment; the second included manually assisted stepping that imposed the periodic load and rhythmic sensory information associated with alternating weight bearing during stepping.
Combined Sensory Input
Despite the small sample size, the immediate stepping responses of the infants in this study suggest that the robust contextual enhancement afforded with a friction treadmill belt4 can be boosted with the addition of a load to the infant's lower legs (F+L), in infants with lumbosacral level lesions. We propose that the added load augmented the enhancement of the friction belt in both stance and swing, given that load alone was unimpressive. During stance, the friction belt alone enhances the likelihood that the treadmill belt will hold and transport the limb further into hip extension, and the load applied to the lower leg then may additionally increase the duration of foot contact with the treadmill belt surface. The increased hip extension then likely enhances the pendular aspects of swing. The finding that the infants in this study took more steps when a load was added to their lower legs, however, supports that these infants were also able to actively participate in swing, including vertical clearance of their feet (to successfully advance a limb over a high friction surface). The additional proprioceptive information from the F+L combination of sensory input may have contributed to a critical threshold of spatiotemporal sensory information necessary to support the coordination of immediate stepping in these infants.7 Indeed, increased extension at the end of stance and an accentuated swing phase with increased vertical clearance are similar to the changes seen in the stepping of spinalized animals after extensive training.13 Each of these 3 components seems to be both assisted and cued with this combination of sensory inputs. This proposed mechanism for the observed effect of the F+L condition requires further study.
Manual provision of alternating cycles of stepping is the primary method of intensive locomotor training for individuals with traumatic SCI.7,11,14,15,18–22 This rehabilitation approach reflects our understanding that task-specific, complex, temporal patterns of sensorimotor information are required to promote plasticity.11 Although adults with SCI show improvements in gait when manual assistance is provided at faster speeds,20 the infants in our study showed an immediate increase in the complexity and frequency of stepping only at the slower speed. These results may reflect that we examined this response in real time, immediately following a trial with provision of manual assistance rather than over time as in intervention/training studies. Still, these preliminary results suggest that manual assistance of stepping may provide an option for promoting an increase in the complexity of stepping in infants with SB, which warrants additional study.
Clinical application of treadmill intervention for infants with SB is still premature. This study did, however, replicate some findings of previous studies that may start to inform the next steps in this area of study. Research in this area suggests that infants with low lumbar lesions may be most responsive to sensory enhancements to boost step frequencies.3,4 Our limited data in this study seem to support this; although the infants with sacral level lesions took more steps on average than infants with lumbar level lesions in the nonenhanced condition, those with sacral level lesions were not the highest responders to the sensory augmentation. Future studies should include detailed sensory and muscle testing of the infants to characterize the infants beyond just lesion levels to further examine this finding. Although the complete and analyzable data from this study were small across all 3 sets of conditions, both the F+L and the manual assistance conditions that include sensory inputs thought to be key elements in locomotor retraining15 in adults with SCI and in spinalized animals, also augmented the immediate step responsiveness of these infants. Further research to examine early facilitation of leg activity in infants with SB is warranted.
An interesting finding in this study was that the 3 infants who had had shunt revisions were the 3 lowest responders to the combined sensory augmentation during treadmill stepping, despite lesion level (Table 1). An underlying hypothesis with this line of research was that enhanced afferent input is necessary to affect residual neural pathways for the infant with greater neurologic involvement. If similar results were found with a larger population of infants, we might refine this overarching hypothesis to separate neurologic involvement of the lower extremities from associated cortical involvement, understanding that cortical involvement (such as that with shunt revisions) may require different input for the same level of step responsiveness. Interestingly, the infants with shunt revisions in this study were the highest responders to optic flow (Table 1 and Figure 3). Only shunt status was reported in Pantall et al,4 limiting the ability to examine the saliency of visual flow in that study by the number of infants with shunt revisions. Because the infants reported in this study were included in that multisite study, it is feasible that the incidence of shunt revisions in the older group may have affected the outcome in favor of visual flow. Further research with a larger sample of infants with SB who have had shunt revisions may be warranted. Given the relationship of the number of shunt revisions to incidence of visuomotor disability later in childhood,23,24 consideration should be given to potential nongait outcomes from early intervention that would include temporally coupled visuomotor input, such as optic flow during stepping.8
The greatest limitation of this study was the small number of infants from a population of infants with highly variable developmental trajectories, lesion level, and associated conditions. Nonparametric analyses with very small sample sizes, however, are considered as powerful as parametric equivalents, and this study used α < .05 to test significance in contrast to α < .10 used in previous studies with slightly larger samples and parametric analyses.3,4 The within-subjects design was both a strength and a limitation. Within-subjects designs reduce the error variance related to individual differences that would otherwise be a concern in a between-subjects design. The limitation specific to this study, however, was that each laboratory visit introduced a new set of conditions, and although the order of the sets of conditions was randomly assigned to control for development over the 3 weeks that could affect immediate step responsiveness, the week-to-week variability in infant behavior was not controllable by averaging data across time.
The generalizability of these findings remains limited by the preliminary stages of this line of research. In these exploratory studies, we are merely determining to what stimuli infants respond, and we look for that response in a rapid and short time frame. Intervention operates on a different time scale, and studies examining training outcomes are necessary before these sensory conditions should be considered for clinical practice. We believe that the next set of questions in this line of inquiry should examine infant responses over time in a training paradigm using what appear to be optimal conditions.4 Furthermore, the characteristics of the infants studied (shunt revision histories, etc...) should be thoroughly described.
This study extended previous work examining the responsiveness of infants with SB to the treadmill context and to sensory enhancements within this context. Unique to this study was the examination of combined sensory input and manually assisted stepping across one group of infants. This study provided further support for the robustness of a high friction treadmill belt with this population of infants and showed that F+L may be even more salient than F alone. This study also showed that manual assistance at a comfortable speed for the infants may be a viable option for increasing the complexity of stepping in real time. In addition, this study raised a question of whether optic flow to enhance stepping may be most effective for those infants with more cortical vulnerability as a consequence of shunt revisions. Further study with larger samples and a smaller set of conditions are necessary next steps.
We thank the infants and families who participated in this study. We also thank Kathleen Sawin, PhD, CPNP-PC, FAAN, and the Spina Bifida Clinic at Children's Hospital of Wisconsin for support with recruitment; Jeff Konrad, Kelly Lynett, and Mina Saeed for assistance with data collection and behavior coding; and Carolyn Heriza for her formative comments on this manuscript. In addition, we thank Beverly D. Ulrich, PhD, University of Michigan, for funding and collaboration.
1. Ulrich D, Ulrich B, Angulo-Kinzler R, Yun J. Treadmill training of infants with Down syndrome: evidence-based developmental outcomes. Pediatrics. 2001;108(5):1–7.
2. Thelen E, Ulrich B. Hidden skills: a dynamic systems analysis of treadmill stepping during the first year. Monogr Soc Res Child Dev. 1991;56(1):1–98.
3. Teulier C, Smith B, Kubo M, et al. Stepping responses of infants with myelomeningocele when supported on a motorized treadmill. Phys Ther. 2009.
4. Pantall A, Teulier C, Smith B, Moerchen V, Ulrich B. Impact of enhanced sensory input on treadmill step frequency: infants born with myelomeningocele. Pediatr Phys Ther. 2011;23(1):42–52.
5. Sival D, van Weerden T, Vles J, et al. Neonatal loss of motor function in human spina bifida aperta. Pediatrics. 2004;114(2):427–434.
6. Geerdink N, Pasman J, Roeleveld N, Rotteveel J, Mullaart R. Responses to lumbar magnetic stimulation in newborns with spina bifida. Pediatr Neurolo. 2006;34(2):101–105.
7. Edgerton V, Roy R. Activity-dependent plasticity of spinal locomotion: implications for sensory processing. Exerc Sport Sci Rev. 2009;37(4):171–178.
8. Moerchen V, Saeed M. Infant visual attention and step responsiveness to optic flow during treadmill stepping. Infant Behav Dev. 2012;37:711–718.
9. Ulrich B, Ulrich D, Angulo-Kinzler R. The impact of context manipulations on movement patterns during a transition period. Hum Mov Sci. 1998;17:327–346.
10. Rossignol S, Frigon A, Barriere G, et al. Chapter 16—Spinal plasticity in the recovery of locomotion. Prog Brain Res. 2011;188:229–241.
11. Harkema S. Neural plasticity after human spinal cord injury: application of locomotor training to the rehabilitation of walking. The Neuroscientist. 2001;7(5):455–468.
12. Bayley N. Bayley Scales of Infant and Toddler Development. 3rd ed. San Antonio, TX: Pearson; 2005.
13. de Leon R, Hodgson J, Roy R, Edgerton V. Retention of hindlimb stepping ability in adult spinal cats after the cessation of step training. J Neurophysiol. 1999;81:85–94.
14. de Leon R, Roy R, Edgerton V. Is the recovery of stepping following spinal cord injury mediated by modifying existing neural pathways by generating new pathways? A perspective.Phys Ther. 2001;81(12):1904–1911.
15. Behrman A, Harkema S. Locomotor training after human spinal cord injury: a series of case studies. Phys Ther. 2000;80:688–700.
16. Chapman D. Context effects on the spontaneous leg movements of infants with spina bifida. Pediatr Phys Ther. 2002;14:62–73.
17. Ivanenko Y, Dominici N, Cappellini G, Lacquaniti F. Kinematics in newly walking toddlers does not depend upon postural stability. J Neurophysiol. 2005;94:754–763.
18. Sadowsky C, McDonald J. Activity-based restorative therapies: concepts and applications in spinal cord injury–related neurorehabilitation. Dev Disabil Res Rev. 2009;15:112–116.
19. Edgerton V, de Leon R, Harkema S, et al. Retraining the injured spinal cord. J Physiol. 2001;533(1):15–22.
20. Dobkin B, Apple D, Barbeau H, et al. Weight-supported treadmill vs overground training for walking after acute incomplete SCI. Neurology. 2006;66:484–493.
21. Dobkin B. Functional rewiring of brain and spinal cord after injury: the three Rs of neural repair and neurological rehabilitation. Curr Opin Neurol. 2000;13:655–659.
22. Barbeau H. Locomotor training in neurorehabilitation: emerging rehabilitation concepts. Neurorehabil Neural Repair. 2003;17(1):1–11.
23. Del Bigio M. Neuropathology and structural changes in hydrocephalus. Dev Disabil Res Rev. 2010;16(1):16–22.
24. Vinck A, Nijhuis-van der Sanden M, Roeleveld N, Mullaart R, Rotteveel J, Maassen B. Motor profile and cognitive functioning in children with SB. Eur J Paediatr Neurol. 2010;14(1):86–92.
analysis of variance; disability evaluation; electromyography; exercise test; female; health status indicators; humans; infant; male; meningomyelocele/rehabilitation; pediatrics; sensation/physiology; skeletal muscle; videotape recording; walking/physiology
© 2013 Lippincott Williams & Wilkins, Inc.