Eur J Appl Physiol DOI 10.1007/s00421-015-3158-0
ORIGINAL ARTICLE
Multi‑muscle coordination during a challenging stance Wen‑Chieh Yang1 · Chih‑Hsiu Cheng2,3 · Hsing‑Kuo Wang1 · Kwan‑Hwa Lin1,5 · Wei‑Li Hsu1,4
Received: 18 September 2014 / Accepted: 19 March 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract Purpose This study aimed to identify the muscle synergies during standing under various sensory contexts in healthy young adults. Methods Sixteen healthy young adults participated in this study. The 4-min stance task was conducted under vision (eyes open or eyes closed) and proprioception (standing on ground or narrowed blocks) manipulated contexts. Electromyography (EMG) of 8 muscles around the right side of the trunk and leg were recorded and submitted to principal component analysis (PCA) to extract the muscle synergies. Two-way ANOVA with repeated measures was employed to test the effect of sensory contexts on the muscle synergies. Results PCA extracted three muscle synergies that accounted for the variance of standing EMG, including
W.-C. Yang and C.-H. Cheng have contributed equally to this work. Communicated by Fausto Baldissera. * Wei‑Li Hsu [email protected] 1
School and Graduate Institute of Physical Therapy, National Taiwan University, Taipei, Taiwan
2
Department of Physical Therapy and Graduate Institute of Rehabilitation Science, College of Medicine, Chang Gung University, Taoyuan, Taiwan
3
Healthy Aging Research Center, Chang Gung University, Taoyuan, Taiwan
4
Physical Therapy Center, National Taiwan University Hospital, Taipei, Taiwan
5
Department of Physical Therapy, Tzu Chi University, Hualien, Taiwan
the push-back (composed of medial gastrocnemius, vastus medialis and biceps femoris), push-forward (composed of tibialis anterior and rectus femoris) and proximal mixed (composed of rectus abdominis, rector spinae, rectus femoris and biceps femoris) synergies. Block-standing increased the contribution of the push-back synergy while decreased the contribution of the push-forward synergy. In addition, contribution of the proximal mixed synergy was higher under ground-standing with eyes open than under blockstanding with eyes open. Conclusion Three muscle synergies were identified during standing in healthy young adults, and the synergies were affected by proprioception but not visual disturbance. The push-back and push-forward synergies showed the opposite response to proprioceptive disturbance, which may result from their antagonism role. Whether this control regime is used for elderly adults or patient populations with movement disorder needs to be further investigated. Keywords Muscle synergy · Postural control · Principal component analysis (PCA) · Electromyography (EMG)
Introduction Postural muscles are activated constantly in the human body to overcome the forces of gravity (Neumann 2002). In quiet stance, the primary postural muscles include soleus, quadriceps, gluteus maximum, erector spinae, abdominal muscles and trapezius. The central nervous system (CNS) coordinates numerous muscles in a human body to maintain stance stability. Instead of being controlled individually, studies found that postural muscles are grouped into functional synergies (Robert et al. 2008; Wang et al. 2005; Krishnamoorthy et al. 2003b). For example, the
13
gastrocnemius, hamstrings and erector spinae muscles work together to counter a sudden backward translation of supporting surface (e.g., standing in a decelerating bus) (Horak et al. 1992). Similar phenomenon could be found at the joints level during static or dynamic stance in younger adults, but the coordination patterns of the joints were altered in the aging populations (Hsu et al. 2013, 2014). Therefore, synergies or coordination patterns are the linkages between goal-level control (e.g., maintain balance stability) and the anatomical implementation (e.g., contraction of muscles or movement of joints) (Tresch and Jarc 2009; Latash et al. 2010). Moreover, muscle synergies reveal more information than individual muscles about the control of CNS during functional tasks and are useful for evaluating coordination patterns (Wang et al. 2015). Horak and Nashner (1986) monitored the electromyogram (EMG) of lower extremities during standing with balance perturbation using the floor translation paradigm. The authors found that a group of ventral muscles including tibialis anterior and rectus femoris activated during forward translation, and a group of dorsal muscles including gastrocnemius and hamstrings activated during backward translation (Horak and Nashner 1986). Henry et al. (1998) extended Horak’s work to multidirectional translation and concluded that the contribution of ventral and dorsal muscles varied with the direction of perturbation. For example, the activation of the ventral muscles gradually decreased from forward translation to backward translation, and the activation pattern was inversed in the dorsal muscles. The results indicated that the mixture of dorsal and ventral muscle synergies forms a continuum of postural responses, and the CNS regulates the weighting of the two synergies in a task-dependent manner. Though muscle synergies reveal important information about the control of CNS, it is difficult to identify these synergies from a group of muscles by inspecting the EMGs directly. Principal component analysis (PCA) is a data reduction technique that can be used to investigate the interaction among multiple muscles and extracts the major synergies among muscles coordination (Krishnamoorthy et al. 2003b, 2004; Wang et al. 2005; d’Avella et al. 2006; Vernazza-Martin et al. 2006). PCA extracts few principal components (PCs) as the linear combinations of the original variables, and these PCs could account for most of the variance in the original dataset (Sharma 1995). For example, a data with 10 variables (i.e., 10 dimensions) could be represented by 3 PCs that account for 90 % of the variance in the original dataset. Since each PC is the linear combinations of the original variables, it reveals how the individual variables were integrated. When it comes to a group of muscle activities, PCA can identify the specific combinations during the movement (Krishnamoorthy et al. 2003b).
13
Eur J Appl Physiol
Extracting muscle synergies by PCA has been used in analyzing numerous tasks, such as weight shifting, load releasing and step initiation. (Krishnamoorthy et al. 2003b). PCA is also adopted in one study that monitored the effect of learning on muscle synergies (Asaka et al. 2008). The authors asked the participants to stand on an unstable surface and found that the synergies shifted from a co-contraction pattern to a reciprocal pattern after practice (Asaka et al. 2008). Most studies investigated the muscle synergies during mechanical perturbations induced extrinsically (e.g., floor translation/tilting) or intrinsically (e.g., arm push/pull). It remains unclear how the muscle synergies were regulated under balance-disturbing sensory contexts. Sensory context builds the internal representation of the body orientation in respect of the gravity and the relative relationship between center of mass and base of support (ShumwayCook and Woollacott 2007). Previous study has shown that postural sway increases when standing with sensory inputs diminished or manipulated (Teasdale et al. 1991; Shumway-Cook and Woollacott 2000). Unlike regaining balance from mechanical perturbations, maintaining balance under sensory disturbances is less likely to produce apparent contraction of the muscles, and numerous muscles could be involved during balance recovery. Therefore, we need a sensitive tool such as PCA to investigate the effect of sensory contexts on muscle synergies. The purpose of this study was to identify the muscle synergies during standing under various sensory contexts in healthy young adults. Specifically, we recorded the EMG of the postural muscles during standing under various vision and proprioception contexts, and extracted the muscle synergies using PCA. We hypothesized that the regulation of muscle synergies during a challenging stance would associate with the direction of body sway. In addition, the variance explained by the muscle synergies would decrease when both visual and proprioceptive inputs are manipulated.
Methods Participants Sixteen healthy right-foot-dominated young adults (6 males and 10 females; mean age 23.1 years old; mean weight 59.2 kg; mean height 166.6 cm) without known neurological or musculoskeletal impairments participated in this study. The dominant foot was determined to be the leg used to kick a soccer ball (Hoffman et al. 1998). The inform consent approved by the Institutional Review Board of National Taiwan University Hospital was signed by all participants prior to this study.
Eur J Appl Physiol
of each 4-min trial. The normalized EMG from one representative trial in ground-standing with eyes open is shown in Fig. 2. The ground reaction force during quiet stance was recorded by two force platforms (model OR6-7, AMTI, Watertown, MA, USA) with sampling rate 1000 Hz and then low-pass-filtered (cutoff 15 Hz). The center of pressure (COP) was then computed offline using customized MATLAB (Mathworks, version 7.10.0, USA) scripts. Data analysis
Fig. 1 Experiment setup shows the block-standing condition. The participant stood on the blocks (hatch filled square) placed upon force platform. The dots upon the body mark the placement of wireless EMG electrodes over the right side
Experimental task Participants were asked to stand barefooted with feet shoulder-width apart, arms comfortably rested by the sides of the body, and maintain the stance for 4 min. The stance task was performed under the combination of 2 vision (eyes open or eyes closed) and 2 proprioception (standing on ground or narrow blocks) contexts, arranged in random sequence. The blocks were 20 × 10 × 12 cm (length, width and height, respectively, Fig. 1) in size, placed under bilateral mid-foot (between metatarsal heads and heel pad). It was assumed that the block-standing constrained the magnitude of postural sway in sagittal plane, thus reduced the proprioceptive input from stretch receptors (Benjuya et al. 2004). In the eyesopen contexts, a stationary visual target was placed 1.6 m in front of the participants. Each sensory context requires two successful trials, which was defined as the maintenance of 4-min stance without imbalance or changing foot placement. Instruments Wireless surface EMG electrodes (Trigno, Delsys Inc., Boston, USA) were placed on eight muscles around the right side of the body, including tibialis anterior (TA), medial gastrocnemius (MG), rectus femoris (RF), vastus medialis (VM), biceps femoris (BF), gluteus maximus (GMX), rectus abdominis (RA), and erector spinae (ES). EMGs were sampled at 1000 Hz, and band-pass-filtered (40–500 Hz) with a zero-phase-shift digital filter (Krishnamoorthy et al. 2003a). The filtered EMGs were normalized to z scores and then rectified to allow comparisons across muscles (Gribble and Ostry 1999). The EMG processing was performed on data from 0.5 to 3.5 min, leaving out the first and last 30 s
Changes in the muscle activation associated with the sway of COP in anteroposterior (AP) direction were calculated in two steps. First, the sway of COP was decomposed into forward sways and backward sways by the first derivative of COP position (≥0 for forward, D + 1.25 × σD D indicates the displacement of each backward sway, while ¯ and σD indicate the mean and standard deviation of all D backward sways, respectively. In our unpublished pilot study, this criterion returned about 40–60 sways with largest displacement and was appropriate for extracting muscle synergies. After COP decomposition, the EMG of the 8 muscles during each COP sway was integrated muscle by muscle to generate an n-by-8 integrated EMG (IEMG) matrix. The n rows indicate the number of COP sways, and the 8 columns indicate the 8 muscles. The IEMG matrix represented the changes in muscle activation associated with the backward sway of COP. The IEMG matrix was submitted to PCA to reduce its dimensionality. PCA extracted few PCs from the linear combinations of the original IEMG matrix, and these PCs could account for most of the variance in the original matrix. Therefore, the dimensionality of original IEMG matrix could be represented by fewer PC subspaces. In most trials, over 90 % of variance was accounted by the first three PCs. Therefore, the first three PCs were selected as the muscle synergies for further analysis. The contribution of individual muscle to each PC was evaluated by the absolute value of factor loading. Muscles with factor loading greater than 0.5 were regarded as significant contributors (Krishnamoorthy et al. 2003b). Statistics The effect of different sensory contexts on the muscle synergies (i.e., the first three PCs) was examined by two-way
13
Eur J Appl Physiol
Fig. 2 Normalized EMGs of 8 recorded muscles from a representative trial in ground-standing with eyes-open context. The dataset is presented from 0.5 to 3.5 min of trial
Table 1 Number of times a muscle is seen as part of a synergy
Muscle
Push-back synergy
Push-forward synergy
Proximal mixed synergy
Tibialis anterior Medial gastrocnemius Rectus femoris Vastus medialis Biceps femoris Gluteus maximum Rectus abdominis
16 43 7 18 26 7 9
40 16 23 0 0 16 11
14 13 18 0 17 0 51
9
16
20
Erector spinae
Bold numbers show muscles that were significantly loaded (loading >0.5) more than 16 cases from all 64 trials
(vision and proprioception) ANOVA with repeated measures. Post hoc Tukey’s HSD test was conducted if the main effect of vision, proprioception or their interaction reached significance level (p
ORIGINAL ARTICLE
Multi‑muscle coordination during a challenging stance Wen‑Chieh Yang1 · Chih‑Hsiu Cheng2,3 · Hsing‑Kuo Wang1 · Kwan‑Hwa Lin1,5 · Wei‑Li Hsu1,4
Received: 18 September 2014 / Accepted: 19 March 2015 © Springer-Verlag Berlin Heidelberg 2015
Abstract Purpose This study aimed to identify the muscle synergies during standing under various sensory contexts in healthy young adults. Methods Sixteen healthy young adults participated in this study. The 4-min stance task was conducted under vision (eyes open or eyes closed) and proprioception (standing on ground or narrowed blocks) manipulated contexts. Electromyography (EMG) of 8 muscles around the right side of the trunk and leg were recorded and submitted to principal component analysis (PCA) to extract the muscle synergies. Two-way ANOVA with repeated measures was employed to test the effect of sensory contexts on the muscle synergies. Results PCA extracted three muscle synergies that accounted for the variance of standing EMG, including
W.-C. Yang and C.-H. Cheng have contributed equally to this work. Communicated by Fausto Baldissera. * Wei‑Li Hsu [email protected] 1
School and Graduate Institute of Physical Therapy, National Taiwan University, Taipei, Taiwan
2
Department of Physical Therapy and Graduate Institute of Rehabilitation Science, College of Medicine, Chang Gung University, Taoyuan, Taiwan
3
Healthy Aging Research Center, Chang Gung University, Taoyuan, Taiwan
4
Physical Therapy Center, National Taiwan University Hospital, Taipei, Taiwan
5
Department of Physical Therapy, Tzu Chi University, Hualien, Taiwan
the push-back (composed of medial gastrocnemius, vastus medialis and biceps femoris), push-forward (composed of tibialis anterior and rectus femoris) and proximal mixed (composed of rectus abdominis, rector spinae, rectus femoris and biceps femoris) synergies. Block-standing increased the contribution of the push-back synergy while decreased the contribution of the push-forward synergy. In addition, contribution of the proximal mixed synergy was higher under ground-standing with eyes open than under blockstanding with eyes open. Conclusion Three muscle synergies were identified during standing in healthy young adults, and the synergies were affected by proprioception but not visual disturbance. The push-back and push-forward synergies showed the opposite response to proprioceptive disturbance, which may result from their antagonism role. Whether this control regime is used for elderly adults or patient populations with movement disorder needs to be further investigated. Keywords Muscle synergy · Postural control · Principal component analysis (PCA) · Electromyography (EMG)
Introduction Postural muscles are activated constantly in the human body to overcome the forces of gravity (Neumann 2002). In quiet stance, the primary postural muscles include soleus, quadriceps, gluteus maximum, erector spinae, abdominal muscles and trapezius. The central nervous system (CNS) coordinates numerous muscles in a human body to maintain stance stability. Instead of being controlled individually, studies found that postural muscles are grouped into functional synergies (Robert et al. 2008; Wang et al. 2005; Krishnamoorthy et al. 2003b). For example, the
13
gastrocnemius, hamstrings and erector spinae muscles work together to counter a sudden backward translation of supporting surface (e.g., standing in a decelerating bus) (Horak et al. 1992). Similar phenomenon could be found at the joints level during static or dynamic stance in younger adults, but the coordination patterns of the joints were altered in the aging populations (Hsu et al. 2013, 2014). Therefore, synergies or coordination patterns are the linkages between goal-level control (e.g., maintain balance stability) and the anatomical implementation (e.g., contraction of muscles or movement of joints) (Tresch and Jarc 2009; Latash et al. 2010). Moreover, muscle synergies reveal more information than individual muscles about the control of CNS during functional tasks and are useful for evaluating coordination patterns (Wang et al. 2015). Horak and Nashner (1986) monitored the electromyogram (EMG) of lower extremities during standing with balance perturbation using the floor translation paradigm. The authors found that a group of ventral muscles including tibialis anterior and rectus femoris activated during forward translation, and a group of dorsal muscles including gastrocnemius and hamstrings activated during backward translation (Horak and Nashner 1986). Henry et al. (1998) extended Horak’s work to multidirectional translation and concluded that the contribution of ventral and dorsal muscles varied with the direction of perturbation. For example, the activation of the ventral muscles gradually decreased from forward translation to backward translation, and the activation pattern was inversed in the dorsal muscles. The results indicated that the mixture of dorsal and ventral muscle synergies forms a continuum of postural responses, and the CNS regulates the weighting of the two synergies in a task-dependent manner. Though muscle synergies reveal important information about the control of CNS, it is difficult to identify these synergies from a group of muscles by inspecting the EMGs directly. Principal component analysis (PCA) is a data reduction technique that can be used to investigate the interaction among multiple muscles and extracts the major synergies among muscles coordination (Krishnamoorthy et al. 2003b, 2004; Wang et al. 2005; d’Avella et al. 2006; Vernazza-Martin et al. 2006). PCA extracts few principal components (PCs) as the linear combinations of the original variables, and these PCs could account for most of the variance in the original dataset (Sharma 1995). For example, a data with 10 variables (i.e., 10 dimensions) could be represented by 3 PCs that account for 90 % of the variance in the original dataset. Since each PC is the linear combinations of the original variables, it reveals how the individual variables were integrated. When it comes to a group of muscle activities, PCA can identify the specific combinations during the movement (Krishnamoorthy et al. 2003b).
13
Eur J Appl Physiol
Extracting muscle synergies by PCA has been used in analyzing numerous tasks, such as weight shifting, load releasing and step initiation. (Krishnamoorthy et al. 2003b). PCA is also adopted in one study that monitored the effect of learning on muscle synergies (Asaka et al. 2008). The authors asked the participants to stand on an unstable surface and found that the synergies shifted from a co-contraction pattern to a reciprocal pattern after practice (Asaka et al. 2008). Most studies investigated the muscle synergies during mechanical perturbations induced extrinsically (e.g., floor translation/tilting) or intrinsically (e.g., arm push/pull). It remains unclear how the muscle synergies were regulated under balance-disturbing sensory contexts. Sensory context builds the internal representation of the body orientation in respect of the gravity and the relative relationship between center of mass and base of support (ShumwayCook and Woollacott 2007). Previous study has shown that postural sway increases when standing with sensory inputs diminished or manipulated (Teasdale et al. 1991; Shumway-Cook and Woollacott 2000). Unlike regaining balance from mechanical perturbations, maintaining balance under sensory disturbances is less likely to produce apparent contraction of the muscles, and numerous muscles could be involved during balance recovery. Therefore, we need a sensitive tool such as PCA to investigate the effect of sensory contexts on muscle synergies. The purpose of this study was to identify the muscle synergies during standing under various sensory contexts in healthy young adults. Specifically, we recorded the EMG of the postural muscles during standing under various vision and proprioception contexts, and extracted the muscle synergies using PCA. We hypothesized that the regulation of muscle synergies during a challenging stance would associate with the direction of body sway. In addition, the variance explained by the muscle synergies would decrease when both visual and proprioceptive inputs are manipulated.
Methods Participants Sixteen healthy right-foot-dominated young adults (6 males and 10 females; mean age 23.1 years old; mean weight 59.2 kg; mean height 166.6 cm) without known neurological or musculoskeletal impairments participated in this study. The dominant foot was determined to be the leg used to kick a soccer ball (Hoffman et al. 1998). The inform consent approved by the Institutional Review Board of National Taiwan University Hospital was signed by all participants prior to this study.
Eur J Appl Physiol
of each 4-min trial. The normalized EMG from one representative trial in ground-standing with eyes open is shown in Fig. 2. The ground reaction force during quiet stance was recorded by two force platforms (model OR6-7, AMTI, Watertown, MA, USA) with sampling rate 1000 Hz and then low-pass-filtered (cutoff 15 Hz). The center of pressure (COP) was then computed offline using customized MATLAB (Mathworks, version 7.10.0, USA) scripts. Data analysis
Fig. 1 Experiment setup shows the block-standing condition. The participant stood on the blocks (hatch filled square) placed upon force platform. The dots upon the body mark the placement of wireless EMG electrodes over the right side
Experimental task Participants were asked to stand barefooted with feet shoulder-width apart, arms comfortably rested by the sides of the body, and maintain the stance for 4 min. The stance task was performed under the combination of 2 vision (eyes open or eyes closed) and 2 proprioception (standing on ground or narrow blocks) contexts, arranged in random sequence. The blocks were 20 × 10 × 12 cm (length, width and height, respectively, Fig. 1) in size, placed under bilateral mid-foot (between metatarsal heads and heel pad). It was assumed that the block-standing constrained the magnitude of postural sway in sagittal plane, thus reduced the proprioceptive input from stretch receptors (Benjuya et al. 2004). In the eyesopen contexts, a stationary visual target was placed 1.6 m in front of the participants. Each sensory context requires two successful trials, which was defined as the maintenance of 4-min stance without imbalance or changing foot placement. Instruments Wireless surface EMG electrodes (Trigno, Delsys Inc., Boston, USA) were placed on eight muscles around the right side of the body, including tibialis anterior (TA), medial gastrocnemius (MG), rectus femoris (RF), vastus medialis (VM), biceps femoris (BF), gluteus maximus (GMX), rectus abdominis (RA), and erector spinae (ES). EMGs were sampled at 1000 Hz, and band-pass-filtered (40–500 Hz) with a zero-phase-shift digital filter (Krishnamoorthy et al. 2003a). The filtered EMGs were normalized to z scores and then rectified to allow comparisons across muscles (Gribble and Ostry 1999). The EMG processing was performed on data from 0.5 to 3.5 min, leaving out the first and last 30 s
Changes in the muscle activation associated with the sway of COP in anteroposterior (AP) direction were calculated in two steps. First, the sway of COP was decomposed into forward sways and backward sways by the first derivative of COP position (≥0 for forward, D + 1.25 × σD D indicates the displacement of each backward sway, while ¯ and σD indicate the mean and standard deviation of all D backward sways, respectively. In our unpublished pilot study, this criterion returned about 40–60 sways with largest displacement and was appropriate for extracting muscle synergies. After COP decomposition, the EMG of the 8 muscles during each COP sway was integrated muscle by muscle to generate an n-by-8 integrated EMG (IEMG) matrix. The n rows indicate the number of COP sways, and the 8 columns indicate the 8 muscles. The IEMG matrix represented the changes in muscle activation associated with the backward sway of COP. The IEMG matrix was submitted to PCA to reduce its dimensionality. PCA extracted few PCs from the linear combinations of the original IEMG matrix, and these PCs could account for most of the variance in the original matrix. Therefore, the dimensionality of original IEMG matrix could be represented by fewer PC subspaces. In most trials, over 90 % of variance was accounted by the first three PCs. Therefore, the first three PCs were selected as the muscle synergies for further analysis. The contribution of individual muscle to each PC was evaluated by the absolute value of factor loading. Muscles with factor loading greater than 0.5 were regarded as significant contributors (Krishnamoorthy et al. 2003b). Statistics The effect of different sensory contexts on the muscle synergies (i.e., the first three PCs) was examined by two-way
13
Eur J Appl Physiol
Fig. 2 Normalized EMGs of 8 recorded muscles from a representative trial in ground-standing with eyes-open context. The dataset is presented from 0.5 to 3.5 min of trial
Table 1 Number of times a muscle is seen as part of a synergy
Muscle
Push-back synergy
Push-forward synergy
Proximal mixed synergy
Tibialis anterior Medial gastrocnemius Rectus femoris Vastus medialis Biceps femoris Gluteus maximum Rectus abdominis
16 43 7 18 26 7 9
40 16 23 0 0 16 11
14 13 18 0 17 0 51
9
16
20
Erector spinae
Bold numbers show muscles that were significantly loaded (loading >0.5) more than 16 cases from all 64 trials
(vision and proprioception) ANOVA with repeated measures. Post hoc Tukey’s HSD test was conducted if the main effect of vision, proprioception or their interaction reached significance level (p
