Cerebral palsy (CP) describes several conditions characterized by nonprogressive, but often changing, motor impairments resulting from brain injury in early development. CP has been reported as a major cause of childhood disability, affecting 2 to 3 per 1000 live-born infants.1,2 Positive symptoms include spasticity, hyperreflexia, and cocontraction; negative symptoms include muscle weakness, sensory deficits, poor balance, and impaired motor coordination.3 Motor coordination is essential to motor function in children with CP.4,5 To assess motor coordination, clinical selective motor control (SMC) tests, involving the ability to activate muscles in a selected pattern, have been used.4 One study demonstrated that deficits in SMC were the most important determinant of decreased gross motor function in children with CP.5 However, the motor coordination during respective movements cannot be fully quantified using the SMC tests alone, because additional information is necessary. Recently, the behavior of muscle synergies has been widely investigated and may be useful for the quantification of motor coordination.6,7
Muscle synergy is believed to control the range of motion and activities of several muscles that coordinate the degrees of freedom of a movement. Evidence indicates that movements are constructed from basic-building blocks that determine muscle activation patterns.8 A method has been used to assess synergies with correlational analyses of muscle activity between muscle pairs.9 However, this method does not fully describe the behavior of synergies because several muscle activities contribute to motor coordination during movement. In recent reports, nonnegative matrix factorization (NNMF) has been used as a method to quantify the behavior of muscle synergies.10
Studies assessing muscle synergies with NNMF have supported that muscle synergies help activate multiple muscles with modular organization. The neural basis underlying the modular organization of muscle activities is encoded in the spinal cord.11,12 Clinical research supports that these neural circuits undergo changes in patients with neural disorders.13
NNMF is a form of factor analysis that determines the number and structure of synergies extracted from electromyography (EMG) data recorded during movement. A previous study suggested that NNMF is distinguished from other methods by a unique iteration algorithm that allows only additive, not subtractive, combinations.14 NNMF is particularly applicable to muscle synergies because of the following 2 properties: firing rates of neurons are never negative, and synapses have an either excitatory or inhibitory function, but do not change sign.14
Using NNMF, the muscle synergies during gait have been demonstrated in people with neural disorders and in infants. Previous studies demonstrated that neural disorders including stroke impact the number and structure of muscle synergies.13,15 One previous study supported that the number of synergies in patients after stroke was decreased compared with that of adults without stroke, and that the number of synergies was significantly associated with gait speed.13 Another study demonstrated that the number of muscle synergies during gait increased from 2 to 4 in the early stages of typical development (TD).16 These studies suggest that the number of synergies detected by NNMF may represent the characteristics of motor coordination in children with CP. However, the understanding of muscle synergies in children with CP remains limited. Furthermore, in a previous study, SMC was defined as the ability to isolate the activation of muscles in a selected pattern in response to the demands of a voluntary movement or posture.17 Therefore, the impairment of SMC would be reflected in a reduction in the number of synergies. We predict that our results will identify the mechanism underlying disorders of SMC, and expand the understanding of the pathological condition and the appropriate choice of therapy.
The aim of this study was to compare the number of synergies during gait between children with CP and children with TD, and to clarify whether specific clinical and gait parameters differed according to the number of synergies in children with CP.
Study Design and Setting
Children with CP were recruited from 2 schools in Kyoto, Japan, as a convenience sample. Age-matched children with TD were recruited from a junior high school in Kyoto, Japan. We obtained informed consent from the children and the parents of each child. All procedures were approved by the Ethics Committee of Kyoto University Graduate School and Faculty of Medicine and were consistent with the tenets of the Declaration of Helsinki.
Fifteen children with CP and 10 children with TD were recruited. The inclusion criteria for the children with CP were (1) Gross Motor Function Classification System (GMFCS) levels I to III, (2) the ability to walk in a straight line for at least 5 m, with minimal or no support, (3) the absence of severe epilepsy, and (4) the absence of seizures during the measurements. Two children with CP who required more than minimal support to walk were excluded. Thirteen children aged 6 to 18 years with CP met the inclusion criteria and participated in the study. Subjects' characteristics are shown in Table 1.
TABLE 1 -
General Characteristics of Children With CP and Children With TD
||Children With CP n = 13 Mean (SD)
||Children With TD n = 10 Mean (SD)
Abbreviations: CP, cerebral palsy; GMFCS, Gross Motor Function Classification System; SD, standard deviation; TD, typical development.
Experimental Protocol and EMG Recordings
Children walked on a 5-m walkway with muscle activity recorded with surface EMG (sEMG) using a Trigno Wireless System (Delsys Inc, sampling rate: 2000 Hz), which recorded data from a 3D accelerometer (ACC). The sEMG activity was recorded from the tibialis anterior, the lateral gastrocnemius, the soleus, the gluteus medius, the rectus femoris, the vastus medialis, the biceps femoris, and the semitendinosus muscles of the most affected side. The ACC was placed on the heel of the measured limb. Furthermore, foot pressure during gait was measured using the FDM system (Zebris Medical GMbh Inc, Germany).
The corrected sEMG data were band-pass filtered (20–250 Hz), rectified, and then low-pass filtered (10 Hz). The recording procedures were similar to those of a previous study.18 Each gait cycle was detected by ACC data on the heel and normalized to 200 data points. Their amplitude was normalized to the peak activity recorded in 5 gait cycles. A factor analysis was performed with the amplitude-normalized and time-normalized sEMG data (nEMG) recorded from each muscle.
Muscle Synergies Extraction
Figure 1 is a schematic of muscle synergy extraction. For each subject, the nEMG data were decomposed to patterns of synergies and their respective muscle weightings using an NNMF algorithm.14 The nEMG data m(t) are represented by the following equation:
This algorithm could reveal synergies in the following 2 matrices: the Ci(t) were the specified activation patterns of each synergy over 5 gait cycles (n × t matrix; n = number of synergies, t = time point: 5 gait cycles × 200 data points), whereas the Wi represented the weightings of the muscles involved in each synergy (m × n matrix; m = 8 muscles in lower limb) and the e represented the residual error. The NNMF algorithm was initialized with 2 random matrices of activation patterns and weightings. For the reconstruction of the nEMG data, the values of these matrices were iteratively updated until the matrices converged.
Determining the Number of Synergies for Each Subject and Group Classification
The NNMF analyses were performed with the outputs, which were constrained from 1 to 8 synergies without a priori assumption. In the course of muscle synergy extraction, the reconstructed EMG (rEMG) was then calculated by performing matrix multiplication with both matrices, and the sum of squared errors (nEMG − rEMG) was also calculated. The variability accounted for (VAF), which is the ratio of the sum of the squared error to the sum of the squared nEMG, was calculated to determine the minimum number of synergies that corresponded to adequate rEMG.13 However, the threshold of VAF had the possibility to change the number of synergies. Thus, we decided that, if the VAF including all muscles was 90% or more, additional synergies were not required, as reported in a previous study.13 The children were then classified into 2 groups corresponding to the number of synergies obtained. The “smaller number of synergies group” (SN) included children who had 2 synergy patterns, whereas the “larger number of synergies group” (LN) included children who had 3 or more synergy patterns.
The modified Ashworth Scale (MAS) was used to assess the degree of spasticity in the knee flexors, ankle dorsiflexors, and ankle plantar flexors; these 3 scores were summed. To ensure the reliability of the measurement, the MAS was measured by the same therapist in all patients.
The modified Trost test of SMC (mT-SMC) was used to measure disorders in SMC.4 The mT-SMC score ranges from 0 to 2, corresponding to the disorder of isolated movements. In the present study, hip flexion, hip abduction, and knee extension movements were assessed. The Spinal Alignment and Range of Motion Measure was used to estimate the extent of deformity and contracture in the lower limb on the most affected side.19 The Pediatric Evaluation of Disability Inventory was used to assess an individual's ability to perform activities of daily living. Gait speed was measured with a stopwatch over a predetermined walking distance. For the biomechanical measures, the position of the center of pressure (COP) was measured and calculated from the displacement of the center of pressure (dCOP) on the measured side during 3 stance phases. These data were subsequently averaged and normalized by lower limb length. The dCOP showed the length of the progression of COP, which relates postural control to gait instability in children with CP.20
The χ2 test was used to compare the number of synergies between children with CP and children with TD. The same test was used to compare the GMFCS level among children with CP. Furthermore, the Mann-Whitney U test and the t test were used to perform comparisons of clinical measures, gait speed, and dCOP between the SN and LN.
Differences in the Number of Synergies Between Children With CP and Children With TD
All children with CP and TD completed the measurements. Figure 2 graphs the distribution in the number of synergies required for adequate reconstruction of muscle activity in children with CP and in children with TD. Among the children with CP, 6 had 2 synergies, 4 had 3 synergies, and 2 had 4 synergies. In contrast, only 1 child with TD had 3 synergies, whereas the remaining children with TD each had 4 or 5 synergies. There was a significant difference in the number of synergies between children with CP and TD (χ2 = 13.64, P = .03). In the children with CP, the number of synergies did not differ on the basis of the GMFCS level (χ2 = 4.06, P = .40).
Differences in Clinical Measurements and Gait Performance Based on the Number of Synergies
Table 2 includes the clinical measures and gait performances in children with CP depending on the number of synergies identified. There were no significant differences between groups in terms of the mT-SMC and Pediatric Evaluation of Disability Inventory scores or gait speed (P = .26, P = .84, and P = 0.14, respectively).
TABLE 2 -
Comparison of Measurements Between 2 Groups Classified by the Number of Synergies in Children With CP
||SN Median or Mean (SD)
||LN Median or Mean (SD)
Abbreviations: CP, cerebral palsy; dCOP, displacement of the center of pressure; LN, larger number of synergy group; MAS, modified Ashworth Scale; mT-SMC, modified Trost test of selective motor control; PEDI, Pediatric Evaluation of Disability Inventory; SAROMM, Spinal Alignment and Range of Motion Measure; SD, standard deviation; SN, smaller number of synergy group.
aSignificant at P < .05.
However, the MAS and Spinal Alignment and Range of Motion Measure scores were significantly different between the SN and LN (P < .05 and P < .01, respectively). Furthermore, the dCOP was significantly greater in the LN than in the SN (P < .05).
The number of synergies during gait was fewer in children with CP than in children with TD. Furthermore, the extent of spasticity significantly differed according to the number of synergies, and the displacement of COP was significantly greater in the group with the larger number of synergies.
Reports regarding the number of synergies in other populations with neurological disorders have been inconsistent. One study demonstrated that the number of synergies in patients after stroke was lower than that in adults without stroke13; however, other studies showed that the number of muscle synergies identified during gait in patients after stroke or spinal cord injury was similar to the number of synergies in adults without neurological insults.10,21 In the present study and a previous study,18 similar findings showed that children with CP had a fewer number of synergies. These results suggest that a decreased number of synergies were observed only in children with CP who experienced a brain injury during early development. The time frame of brain injury is thought to play an important role in establishing the differing characteristic between children with CP and patients after stroke or spinal cord injury. Children with CP suffered a brain injury before the acquisition of independent gait. There is also the possibility that the number of synergies reflects changes in the neural system because of development arising from the experience of independent gait.
Dominici et al16 demonstrated that 2 basic synergy patterns were retained and tuned, whereas 2 new patterns were added in the beginning of walking independently in early development. That is, in infants without neurological insult, the number of synergies gradually increases during motor development. The present study showed significant differences in the extent of spasticity depending on the number of synergies in children with CP, suggesting that severe spasticity, resulting from severe brain injury, limits the increases in the number of synergies during development in children with CP.
The association between the number of synergies and the gait performance has been demonstrated in patients after stroke.13 In children with CP, the dCOP has been investigated as a general measure of postural control and gait instability, and the dCOP has been observed to be shorter in children with CP than in children with TD.20 In the present study, the dCOP during gait significantly varied on the basis of the number of synergies, which suggests that the number of synergies is a critical determinant of gait performance in children with CP.
With regard to the clinical importance of the number of synergies, a previous study demonstrated that the number of synergies is associated with gait speed in patients with central nervous system disorders.13 Changes in primary gait patterns may impact future gait performance in children with CP. Because earlier reports have suggested that muscle synergies are related to biomechanical output,6 it is expected that changes in muscle synergies would directly contribute to disorders of biomechanical output.
A previous study explained the mechanism underlying increases or decreases in the number of synergies, using the merging or fractionation of muscle synergies in patients with neural disorders.15 Moreover, the clinical factors that are related to the merging and fractionation of muscle synergies in patients after stroke were identified in one of our previous studies.22 Indeed, one of our findings demonstrated that the merging of muscle synergies was dependent upon the restriction of an increase in muscle strength, whereas other findings showed that the fractionation of muscle synergies was associated with an improvement in activities of daily living. Therefore, a reduction in the number of synergies in children with CP might represent one of the neural adaptations following the improvement of motor function. Our previous results also suggest that therapists should provide significant movement experiences in activities of daily living for children with CP starting in early childhood.
The present study did not reveal differences in gait speed according to the number of synergies, which may be due to the following limitations of the present study. First, the sample size was small and yielded a limited variability in motor function among the subjects. Second, children who required minimal assistance (GMFCS III) were included, and the type of assistance used during gait may have influenced the comparison of gait speeds.
Furthermore, in the initial phases of gait development, and after gait development, the changes in the number of synergies have not yet been fully clarified. Therefore, future longitudinal studies are required to demonstrate the changes in synergies following the recovery of motor function in children with CP.
The primary finding of the present study is that the extent of spasticity and gait kinetics differed depending on the number of synergies during gait in children with CP. Our results showed that muscle synergies during gait were significantly associated with impairment and gait performance in children with CP. Furthermore, the NNMF has the potential to enable the quantified assessment of motor coordination in children with CP.
1. Yeargin-Allsopp M, Van Naarden Braun K, Doernberg NS, Benedict RE, Kirby RS, Durkin MS. Prevalence of cerebral palsy in 8-year-old children in three areas of the United States in 2002: a multisite collaboration. Pediatrics. 2008;121(3):547–554.
2. Himmelmann K, Uvebrant P. The panorama of cerebral palsy in Sweden. XI. Changing patterns in the birth-year period 2003-2006. Acta Paediatrica. 2014;103(6):618–624.
3. Kerr Graham H, Selber P. Musculoskeletal aspects of cerebral palsy. J Bone Joint Surg Br. 2003;85(2):157–166.
4. Smits DW, van Groenestijn AC, Ketelaar M, Scholtes VA, Becher JG, Gorter JW. Selective motor control of the lower extremities in children with cerebral palsy: inter-rater reliability of two tests. Dev Neurorehabil. 2010;13(4):258–265.
5. Voorman JM, Dallmeijer AJ, Knol DL, Lankhorst GJ, Becher JG. Prospective longitudinal study of gross motor function in children with cerebral palsy. Arch Phys Med Rehabil. 2007;88(7):871–876.
6. d'Avella A, Saltiel P, Bizzi E. Combinations of muscle synergies in the construction of a natural motor behavior. Nat Neurosci. 2003;6(3):300–308.
7. Ivanenko YP, Poppele RE, Lacquaniti F. Motor control programs and walking. Neuroscientist. 2006;12(4):339–348.
8. Bernstein N. The Coordination and Regulation of Movements. Oxford, England: Pergamon Press; 1965.
9. Zwaan E, Becher JG, Harlaar J. Synergy of EMG patterns in gait as an objective measure of muscle selectivity in children with spastic cerebral palsy. Gait Posture. 2012;35(1):111–115.
10. Ivanenko YP, Grasso R, Zago M, et al. Temporal components of the motor patterns expressed by the human spinal cord reflect foot kinematics. J Neurophysiol. 2003;90(5):3555–3565.
11. Hart CB, Giszter SF. A neural basis for motor primitives in the spinal cord. J Neurosci. 2010;30(4):1322–1336.
12. Takei T, Seki K. Synaptic and functional linkages between spinal premotor interneurons and hand-muscle activity during precision grip. Front Comput Neurosci. 2013;7:40.
13. Clark DJ, Ting LH, Zajac FE, Neptune RR, Kautz SA. Merging of healthy motor modules predicts reduced locomotor performance and muscle coordination complexity post-stroke. J Neurophysiol. 2010;103(2):844–857.
14. Lee DD, Seung HS. Learning the parts of objects by non-negative matrix factorization. Nature. 1999;401(6755):788–791.
15. Cheung VC, Turolla A, Agostini M, et al. Muscle synergy patterns as physiological markers of motor cortical damage. Proc Natl Acad Sci U S A. 2012;109(36):14652–14656.
16. Dominici N, Ivanenko YP, Cappellini G, et al. Locomotor primitives in newborn babies and their development. Science. 2011;334(6058):997–999.
17. Sanger TD, Chen D, Delgado MR, Gaebler-Spira D, Hallett M, Mink JW. Definition and classification of negative motor signs in childhood. Pediatrics. 2006;118(5):2159–2167.
18. Li F, Wang Q, Cao S, et al. Lower-limb muscle synergies in children with cerebral palsy. Presented at: 6th International Ieee/Embs Conference on Neural Engineering; 2013:1226–1229.
19. Chen CL, Wu KP, Liu WY, Cheng HY, Shen IH, Lin KC. Validity and clinimetric properties of the Spinal Alignment and Range of Motion Measure in children with cerebral palsy. Dev Med Child Neurol. 2013;55(8):745–750.
20. Hsue BJ, Miller F, Su FC. The dynamic balance of the children with cerebral palsy and typical developing during gait Part II: instantaneous velocity and acceleration of COM and COP and their relationship. Gait Posture. 2009;29(3):471–476.
21. Gizzi L, Nielsen JF, Felici F, Ivanenko YP, Farina D. Impulses of activation but not motor modules are preserved in the locomotion of subacute stroke patients. J Neurophysiol. 2011;106(1):202–210.
22. Hashiguchi Y, Ohata K, Kitatani R, et al. Merging and Fractionation of muscle synergy indicate the recovery process in patients with hemiplegia: the first study of patients after subacute stroke. Neural Plast. 2016;2016:5282957.