Clarification of Muscle Synergy Structure During Standing-up Motion of Healthy Young, Elderly and Post-Stroke Patients

Transcription

1 2017 International Conference on Rehabilitation Robotics (ICORR) QEII Centre, London, UK, July 17-20, Clarification of Muscle Synergy Structure During Standing-up Motion of Healthy, Elderly and Post- N. Yang 1, Q. An 1, H. Yamakawa 1, Y. Tamura 1, A. Yamashita 1, K. Takahashi 2, M. Kinomoto 2, H. Yamasaki 3, M. Itkonen 3, F. S. Alnajjar 3, S. Shimoda 3, H. Asama 1, N. Hattori 2 and I. Miyai 2 Abstract Standing-up motion is an important daily activity. It has been known that elderly and post-stroke patients have difficulty in performing standing-up motion. The standing-up motion is retrained by therapists to maximize independence of the elderly and post-stroke patients, but it is not clear how the elderly and post-stroke patients control their redundant muscles to achieve standing-up motion. This study employed the concept of muscle synergy to analyze how healthy young adults, healthy elderly people and post-stroke patients control their muscles. Experimental result verified that four muscle synergies can represent human standing-up motion. In addition, it indicated that the post-stroke patients shift the weights of muscle synergies to finish standing-up motion comparing to healthy subjects. Moreover, different muscle synergy structures were associated with the CoM and joint kinematics. I. INTRODUCTION Since 1950, the world s elderly population has been increasing sharply. According to United Nation, the proportion of the world s population aged 60 years or over increased from 8% in 1950 to 12% in 2013, and it is estimated that it will increase more rapidly to reach 21% in 2050 [1]. In order to improve the quality of life of the elderly, the standingup motion is focused as one of the most common daily activities, which usually followed by other activities. For the elderly with lower extremity muscle weakness, it is difficult for them to rise from a chair. Alexander et al. [2] found that the elderly generated less knee extension torque than young adults. However, the elderly need to use more knee extension torque to stand-up from a chair than that of young adults. Muscle strength was further impaired in post-stroke patients comparing to the elderly [3]. It is also found that compared to the healthy elderly people, the post-stroke patients need more time to do standing-up motion [4][5]. Hence, the standing-up motion is retrained by therapists to maximize independence of the elderly and post-stroke patients. To retrain the elderly and the post-stroke patients efficiently, it is important to understand the mechanism of the standing-up motion. Hughes et al. defined three strategies (momentum tansfer, stabilization, and hybrid) used in humans standing-up motion which generate different movements [6]. They found that younger persons tend to use the 1 N. Yang, Q. An, H. Yamakawa, Y. Tamura, A. Yamashita, and H. Asama are with Department of Precision Engineering, Graduate School of Engineering, The University of Tokyo, Tokyo, Japan yang@robot.t.u-tokyo.ac.jp 2 K. Takahashi, M. Kinomoto, N. Hattori, and I. Miyai are with Morinomiya Hospital, Osaka, Japan 3 H. Yamasaki, M. Itkonen, F. S. Alnajjar, and S. Shimoda are with RIKEN Brain Science Institute, Japan momentum transfer strategy which utilizes the momentum to lift up their body. On the other hand, the elderly persons tend to use the stabilization strategy in which they carry their center of mass (CoM) first on their feet and then they move upward. The hybrid strategy is the middle between momentum transfer and stabilization. Other studies also found that post-stroke patients have delayed muscle activation compared to healthy elderly people [7]. However, they could not reveal how humans coordinate muscles to generate different standing-up movements. Essentially human movement is varied because human body is a redundant system that there are more muscles than the number of joints to be controlled. In order to clarify how humans coordinate their redundant number of muscles to generate different standing-up movements, the concept of muscle synergy has been employed in the area of human motor control theory. Muscle synergy was firstly proposed by Bernstein, which suggested that human movements could be generated from a limited number of modules (called muscle synergy) [8]. It claimed that humans did not control individual muscles, but they controlled modules of muscles. Muscle synergy theory decomposes the complex control of individual muscles into modular organization. Previous studies showed human locomotion could be explained by a small number of muscle synergies [9][10]. Ivanenko et al. found that muscle synergy structure were similar when healthy humans walked with different speeds and gravitational loads, but they adaptively changed the weighting coefficient of muscle synergy to achieve the different locomotion [9]. These findings suggested that humans might utilize different combinations or different ways to activate the limited number of muscle synergies to accomplish adaptive movements. For human standing-up movements, both forward dynamic simulation and measurement experiment were employed to clarify the muscle synergy structure in our previous studies [11]. However, the measurement experiment was based on healthy young adults. It is still unclear how the healthy elderly people and post-stroke patients coordinate redundant muscles to achieve standing-up motion. Therefore, we perform the standing-up motion experiment of healthy elderly people and post-stroke patients in this study. We aim at clarifying the muscle synergy structure in standingup motion of the healthy elderly people and post-stroke patients. The differences among the healthy humans and post-stroke patients were compared. In addition, muscle synergy structure of the post-stroke patients were especially focused /17/$ IEEE 19

2 II. METHOD A. Subjects In this experiment, six healthy young adults (25±3.0 years old), five healthy elderly participants (66.8±8.5 years old) and three post-stroke patients (58.3±11.4 years old) were asked to stand up from their own comfortable feet location. Three months have passed after the stroke for two patients, and five months have passed for one subject. All of the subjects could stand up from a chair by themselves without any support. Average Fugl-Meyer Assessment score was 72.7±6.6. There were 15 trials for each healthy young and elderly participants, and 10 trials for each post-stroke patients in the experiment. The chair height was adjusted to the height of lower leg. The participants finished the motion without moving their feet in all the trials. The informed consent was obtained from all participants according to the protocol of the Institute Review Board of The University of Tokyo and Morinomiya Hospital, Japan. B. Experiment Setting In this experiment, the motion capture system (Motion Analysis. Corp.) was used to record the movements of joints in 100 Hz. This motion capture system has eight infrared cameras. Human joint positions were measured based on Helen Hayes marker set. Joint angle data was calculated using SIMM (Musculographics, Inc) based on the joint position data and the jiont angle data was used to calculate the trajectories of CoM. Surface EMG device (DL-141, S&ME Corp.) was used in this experiment to get muscle activities data in 2,000 Hz. Ten muscles related to standing up motion were measured according to their contribution to extend or flex the ankle, knee, and hip joints: tibialis anterior (TA), gastrocnemius (GAS), soleus (SOL), rectus femoris (RF), vastus lateralis (VAS), biceps femoris long head (BFL), biceps femoris short head (BFS), gluteus maximus (GM), rectus abdominis (RA) and erector spine (ES). The names of the ten muscles are shown in Fig. 1. Two force plate devices (TechGihan. Corp.) were used to record the reaction force data in 2,000 Hz. The force data was used to define the start time of phases. C. Muscle Synergy Model Bernstein suggested that human movement was composed of the small sets of modules (called synergy). It assumes humans do not control individual muscles but they control group of muscles. The muscle synergy model has been widely used to analyze human movements [9][10]. For the muscle synergy model, muscle activation can be expressed as a linear summation of spatiotemporal patterns in a mathematical expression, as in Eq. (1), M = WC, (1) where matrices M, W, and C indicate muscle activation, spatial pattern and temporal pattern matrices respectively. Matrix RA RF VAS TA (a) Front Fig. 1. (b) Back Measured Muscles. ES GMA BFL BFS GAS SOL M consists of the muscle activation vectors m i(i=1,2,...,n) to represent n different muscle activations (Eq. (2)). M = ( m 1 (t) m 2 (t)... m n (t) ) T m 1 (1)... m 1 (t max ) = (2) m n (1)... m n (t max ) The components of the vector are m i (t) to represents discrete i-th muscle activation at time t (1 t t max ). Variable n represents the number of muscles. Spatial pattern W is used to represent the relative activation level of muscle. Variable N indicates the number of muscle synergies. Its column shows N different spatial pattern vectors w j( j=1,2,...,n). The vector w j consists of w i j which represents relative activation level of muscle i included in j-th muscle synergy (Eq. (3)). W = ( ) w 1 w 2... w N = w w 1N..... w n1... w nn. (3) Temporal pattern C is used to indicate the time-varying weighting coefficient of N muscle synergies (Eq. (4)). Its row shows N different temporal pattern vectors c j( j=1,2,...,n), which indicate temporal pattern corresponded to the spatial patterns w j. Its components are c j (t), which represent weighting coefficient of j-th muscle synergy at time t. C = ( c 1 (t) c 2 (t)... c N (t) ) T c 1 (1)... c 1 (t max ) = (4) c N (1)... c N (t max ) Figure 2 shows a schematic design of muscle synergy model. Three muscle synergies are used to express n muscle activations. They are composed of spatial and temporal patterns. Spatial patterns (w 1,2,3 ) show relative muscle activation 20

3 Activation Weight Time [s] Muscle Activation (a) Spatial Pattern (b) Temporal Pattern (c) Muscle Activation Fig. 2. Muscle synergy model. level. Temporal patterns (c 1,2,3 ) show the relative weighting coefficients. During the motion, spatial patterns are constant, but temporal patterns change according to the time. Muscle activations are generated from linear summation of spatial and temporal patterns of muscle synergies. Muscle activations are shown in gray areas and muscle synergies 1, 2, and 3 are described in solid lines, dashed lines and solid lines with circles in 2. To calculate elements of the matrices W and C, nonnegative matrix factorization (NNMF) [12] was used. After obtaining the spatiotemporal patterns of muscle synergy, the determination of coefficient R 2 was calculated to test if the muscle synergy can well represent muscle activation. R 2 was also used in previous studies to analyze the ability to activate muscle synergy independently [14]. This study conducts a measurement experiment of stroke patient performing standing-up motion and it would be analyzed how the muscle synergy structure differs among different types of standingup motion. It was found that there are fewer muscle synergies in the walking of post-stroke patients [10]. The reduction in muscle synergy numbers suggested that post-stroke patients had lower locomotor output complexity which associated with poor walking performance. D. Center of Mass Calculation In order to calculate the center of mass (CoM) position, human body is expressed in four link model: 1. HAT (head, arm and trunk), 2. pelvis, 3. thigh and 4. shank. This study focuses on sagittal movement and horizontal and vertical CoM positions CoM x (t) and CoM y (t) at time t are calculated from Eqs. (7)-(8). In the equations, M k, p x k (t) and py k (t) indicate the mass, horizontal and vertical CoM positions of the k-th segment. p x k (t) and py k (t) are calculated from the ratio of CoM to the segment length and joint angle data of ankle, knee, hip and lumbar was used [15]. CoM x (t) = 4 k=1 M k p x k (t) 4 k=1 M, k (5) CoM y (t) = 4 k=1 M k p y k (t) 4 k=1 M. k (6) E. Kinematic Events of Standing-up Motion In the experiment, the duration time of the standingup motion is different among trials and subjects. In order Fig. 3. Definition of phases. to compare different trials, the duration time needs to be normalized. To normalize the duration time, the standing-up motion is divided into four phases [13], as shown in Fig. 3. III. RESULTS AND DISCUSSION A. Trajectory of Center of Mass Figure 4 shows the trajectory of CoM during the standingup motion. The black solid circled lines, solid lines and dashed lines show the CoM trajectories of the healthy young adults, the elderly and post-stroke patients respectively. The displacements of CoM on both horizontal and vertical directions were normalized based on the height of each subject. The gray part shows the feet location. Comparing to the healthy young adults and the elderly, the post-stroke patients first moved the CoM to their feet before they moved upward. As shown in the result of temporal patterns, the post-stroke patients used longer time to move the CoM forward before extended the body to move upward. This result of CoM also showed that the post-stroke patients firstly moved the CoM to the feet to make the standing-up motion more stable. Figures 4 (a) and (b) also show that healthy young adults and the elderly moved the CoM upward before the CoM reaching the feet position. Therefore, the young and the elderly needed to decelerate the horizontal CoM movement to keep balance at the end of the motion. This is related to the results of temporal pattern of muscle synergy 4. The healthy young adults and the elderly reached the peak time ealier than that of the post-stroke patients. B. The Number of Synergies Figure 5 shows the coefficient of determination R 2 of different numbers of muscle synergies. The black solid circled lines, solid lines and dashed lines respectively represent healthy young adults, the elderly and post-stroke patients. For the healthy young adults and the elderly, the coefficients of determination R 2 were 94±2% and 88±3% when the number of muscle synergies was four. This indicate that four muscle synergies can represent most of the muscle activation during the standing-up motion. In contrast, R 2 was 88±4% for the post-stroke patients and when four muscle synergies were used to represent muscle activation. In addition, more numbers of muscle synergies increased the value of R 2 little. To compare the characteristics of muscle synergy structure between the healthy participants and post-stroke patients, the 21

4 Normalized Vertical Displacement [%] Feet Location Normalized Vertical Displacement [%] Feet Location Normalized Vertical Displacement [%] Feet Location Normalized Horizontal Displacement [%] -5 Normalized Horizontal Displacement [%] -5 Normalized Horizontal Displacement [%] (a) CoM of young adults (b) CoM of the elderly Fig. 4. Trajectory of CoM. (c) CoM of post-stroke patients Coefficient of Determination The Number of Muscle Synergy Coefficient of Determination The Number of Muscle Synergy Coefficient of Determination The Number of Muscle Synergy (a) Healthy young adults Fig. 5. (b) Healthy elderly people Coefficient of determination. (c) Post-stroke patients same number of muscle synergies were also used to represent the muscle activation level of the patients [14]. C. Spatial and Temporal Patterns of Muscle Synergies The result of muscle synergy structure were shown in Fig. 6. Figures 6 (a), (c), (e), and (g) show the spatial patterns of four muscle synergies, which represent muscle activation of the ten selected muscles. Black, white and gray bars show relative muscle activation of healthy young adults, the elderly and the post-stroke patients respectively. Each spatial pattern had particular contribution to body movements according to human anatomy. In Fig. 6 (a), it shows that muscle synergy 1 mostly activated RA which flexed the lumbar and produced momentum necessary for the standing-up motion. Muscle synergy 2 mostly activated TA which dorsiflexed the ankle joint to move center of mass (CoM) forward, and activated VAS and RF to extend the knee as in Fig. 6 (c). In Fig. 6 (e), muscle synergy 3 mainly activated ES, VAS, BFL, BFS to extend trunk and knee, which contribute to extend the whole body. Figure 6 (g) shows that muscle synergy 4 activated GAS, SOL and GMA, which planterflexed the ankle and extend the hip to decelerate movement of CoM and control the standing posture to stay stable. The young, the elderly and the post-stroke patients show the same characteristic muscle activations. This implies that spatial patterns of muscle synergies would be preserved even in the post stroke patients as well. However, further study will be necessary for the wider range of subjects. Current study employed the post stroke patients who comparatively had better motor function (FMA scor is larger than 70), and different spatial patterns can be observed from the patients who had lower mobility. Figures 6 (b), (d), (f) and (h) show the temporal pattern of the four muscle synergies. The vertical solid lines, solid circled lines and dashed lines show the start time of the phases 1 4 of healthy young adults, healthy elderly people and poststroke patients respectively. The start time of each phase is shown in Table I. The duration time of the whole standing-up motion is normalized to 100% from the beginning of phase 1 to the end of phase 4. The solid lines with circle markers, solid lines and dashed lines show the temporal patterns of the elderly, the healthy young and post-stroke patients. In Fig. 6 (b), muscle synergy 1 was firstly activated during the phase 1. It shows that the post-stroke patients had longer time to bend their trunk forward. This difference results in the downward movement of CoM at the beginning of their movement as in Fig. 4 (c) whereas downward movement was not observed for the young and the elderly as in Figs. 4 (a) TABLE I ONSET TIME OF FOUR PHASES. BELOW SHOWS START TIME OF EACH PHASES. phase 2 phase 3 phase 4 Healthy young 33% 46% 80% Healthy old 31% 48% 80% Stroke Patient 38% 65% 80% 22

5 Phase 2 Elderly (a) Spatial pattern of muscle synergy 1 (b) Temporal pattern of muscle synergy 1 Phase 2 Elderly (c) Spatial pattern of muscle synergy 2 (d) Temporal pattern of muscle synergy 2 Elderly Phase 3 (e) Spatial pattern of muscle synergy 3 (f) Temporal pattern of muscle synergy 3 Phase 4 Elderly (g) Spatial pattern of muscle synergy 4 (h) Temporal pattern of muscle synergy 4 Fig. 6. Spatiotemporal patterns of four muscle synergies. and (b). The characteristic difference can be observed in the temporal patterns of muscle synergy 2 (Fig. 6 (d)). This shows that the peak times comes in the order of the young, the elderly, and the stroke patients. Duration time of phase 2 is also the longest for the patients (28%) compared to the young (13%) and the elderly (17%). Similarly the gradient after the peak value differs for three types of the subjects. The patients had the most gentle gradient, the elderly had the second most gentle one, and the young had the most steep gradient. Taking CoM trajectory into account, this phenomena is related to the horizontal CoM movement. At the time of upward movement of CoM, the post-stroke patients moved his CoM closest to the feet. On the other hand, CoM of the young subjects is the farthest from the feet position, and the CoM trajectory of the elderly was the middle between the young and the post-stroke patients. This implies that humans controls the activation timing and duration of the muscle synergy 2 to control horizontal CoM movement. In Fig. 6 (f), temporal patterns of the post-stroke patients differed from the young and the elderly. Since the start time of the phase 3 (body extension phase) is the latest for the post-stroke patients, the peak time was delayed accordingly. As it has been discussed above, the patients firstly move the CoM on the feet and start moving upward. Delayed activation of the muscle synergy 3 resulted in the late upward 23

6 movement. At last, the muscle synergy 4 was activated as in Fig. 6 (h). This muscle mainly activate the muscle in the back side of human body (GAS, SOL, BFL, BFS, GMA) to decelerate the horizontal movement. As well as other synergies, the gradient became more gentle in the order of the poststroke patients, the elderly, and the young. This phenomena could be interpreted as the compensation of the forward horizontal movement of the muscle synergy 2. When the muscle synergy 2 activated latter and its gradient became gentle, the same trend was observed for the muscle synergy 4 accordingly. This compensated activation is considered to be important because this maintains the CoM on their feet. If the post stroke patients cannot control the start time of the muscle synergy 4 properly, the movement would result in falling down. IV. CONCLUSIONS In this study, muscle synergy model was employed to analyze how the healthy young adults, the healthy elderly people and post-stroke patients controlled their muscles to do standing-up motion. Experimental result verified that four muscle synergies could represent human standing-up motion. Our findings showed that the spatial pattern of the muscle synergies were common among the young, the elderly and the patients. However, the post-stroke patients could utilize the same synergies to achieve more stable strategy of standing-up motion in which they firstly move their CoM on their feet and extend their body upward. The results showed the possibility that the humans changed the temporal patterns of the muscle synergy to realize the different strategies of the standing-up motion. In the future study, wider range of post-stroke patients will be studied. If the muscle synergy structure, including peak time or gradient of the temporal patterns, differs according to the severity of the stroke, this can be used for evaluation of the patients, or it can provide useful information for rehabilitation methodology. [5] M. Engardt, Long-term effects of auditory feedback training on relearned symmetrical body weight distribution in stroke patients: a follow-up study, Scandinavian Journal of Rehabilitation Medicine, vol. 26, pp , [6] M. A. Hughes, D. K. Weiner, M. L. Schenkman, R. M. Long, and S. A. Studenski, Chair rise strategies in the elderly, Clinical Biomechanics, vol. 9, pp , [7] P. T. Cheng, C. L. Chen, C. M. Wang and W. H. Hong, Leg Muscle Activation Patterns of Sit-to-Stand Movement in, American Journal of Physical Medicine & Rehabilitation, vol. 83, pp , [8] N. Bernstein, The coordination and regulation of movement, Pergamon, Oxford, [9] Y. P. Ivanenko, R. E. Poppele and F. Lacquaniti, Five basic muscle activation patterns account for muscle activity during human locomotion, Journal of Physiology, vol. 556, pp , [10] D. J. Clark, L. H. Ting, F. E. Zajac, R. R. Neptune and S. A. Kautz, Merging of healthy motor modules predicts reduced locomotor performance and muscle coordination complexity post-stroke, Journal of Neurophysiology, vol. 103, pp , [11] Q. An, Y. Ishikawa, S. Aoi, T. Funato, H. Oka, H.Yamakawa, A. Yamashita and H. Asama, Analysis of muscle synergy contribution on human standing-up motion using human neuro-musculoskeletal model, Proceedings of the 2015 IEEE International Conference on Robotics and Automation (ICRA2015), pp , [12] M. W. Berry, M. Browne, A. N. Lagnville, V. P. Pauca and R. J. Plemmons, Algorithms and applications for approximate nonnegative matrix factorization, Computational Statistics and Data Analysis, vol. 52, pp , [13] M. Schenkman, R. A. Berger, P. O. Riley, R. W. Mann and W. A. Hodge, Whole-body movements during rising to standing from sitting, Physical Therapy, vol. 70, pp , [14] D. J. Clark, L. H. Ting, F. E. Zajac, R. R. Neptune and S. A. Kautz, Merging of healthy modules predicts reduced locomotor performance and muscle coordination complexity post-stroke, Journal of Neurophysiol, vol. 103, pp , [15] N. Ogihara and N. Yamazaki, Generation of human bipedal locomotion by a biomimetic neuro-musculo-skeletal model, Biological Cybernetics, vol. 84, pp. 111, ACKNOWLEDGMENT This work was supported by JSPS KAKENHI Grant Number 15K20956, , 16H04293, JST RISTEX Service Science, Solutions and Foundation Integrated Research Program. REFERENCES [1] United Nations, Department of Economic and Social Affairs, Population Division (2013). World population ageing [2] N. B. Alexander, M. J. Gu, M. Branch, A. B. Schultz, J. A. Ashton- Miller and M. M. Gross, Geriatrics: does leg torque influence rising from a chair in older adults, Rehabilitation Research & Development Program Report, vol. 32, pp , [3] P. H. McCrea, J. J. Eng, A. J. Hodgson, Time and magnitude of torque generation is impaired in both arms following stroke, Muscle & Nerve, vol. 28, pp , [4] S. W. Chou, A. M. Wong, C. P. Leong, W. S. Hong, F. T. Tang and T. H. Lin, Postural control during sit-to-stand and gait in stroke patients, Archives of Physical Medicine and Rehabilitation, vol. 82, pp ,