Journal of Neurological Research And Therapy

Journal of Neurological Research And Therapy

Journal of Neurological Research and Therapy

Current Issue Volume No: 1 Issue No: 4

Research Article Open Access Available online freely Peer Reviewed Citation

Muscle Activation Signals During Gait Parkinson’s Disease are More Rhythmic than in Healthy Controls

1Center for Neuromotor and Biomechanics Research, University of Houston, Houston, TX, USA

2The Methodist Neurological Institute, Houston, TX, USA



To quantify the underlying rhythmic patterns observed in surface electromyography (SEMG) in patients with Parkinson’s Disease (PD), to compare rhythmicity of PD gait with normal gait, and to compare overground walking to treadmill gait.


Eight individuals with idiopathic PD, and ten individuals with no history of neurological disorders participated. SEMG was recorded from the vastus lateralis, tibialis anterior, lateral gastrocnemius and biceps femoris muscles bilaterally. Each participant performed two trials of overground walking and two trials of treadmill walking at a self-selected comfortable walking speed. SEMG was analyzed using a non-linear statistical model that identified underlying recurrent locomotor patterns, which estimated the relative contribution of central pattern generators to the observed muscle activation signals. An index of rhythmicity was determined from the statistic, R2.


The rhythmicity of PD gait was significantly higher than that of normal gait (p = .0458). There was also a significant difference between the rhythmicity of overground walking and treadmill walking (p = .0097).


Individuals with PD appear to walk with muscle activation patterns that are more rhythmic than normal. This suggests that there is more stride-to-stride consistency, and there are fewer postural adjustments and responses to perturbations. We also found that treadmill gait was more rhythmic than overground walking. These findings, although preliminary, challenge the paradigm and current approach to gait retraining of patients with PD.

Author Contributions
Received 28 Sep 2015; Accepted 17 Aug 2016; Published 20 Aug 2016;

Academic Editor: Mourad Tayebi, The University of Surrey

Checked for plagiarism: Yes

Review by: Single-blind

Copyright ©  2016 Timothy Adam Thrasher, et al.

Creative Commons License     This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Competing interests

The authors have declared that no competing interests exist.


Timothy Adam Thrasher, Stanley Fisher (2016) Muscle Activation Signals During Gait Parkinson’s Disease are More Rhythmic than in Healthy Controls . Journal of Neurological Research And Therapy - 1(4):1-9.

Download as RIS, BibTeX, Text (Include abstract )

DOI 10.14302/issn.2470-5020.jnrt-15-750


Parkinson’s Disease (PD) is a complex neurodegenerative disorder that affects approximately 0.3% of the population worldwide1. Clinical management of PD currently focuses on pharmacological treatment and surgically implanted devices2. While these approaches effectively treat some motor symptoms, they are less successful at treating postural instability and gait disturbances, especially as the disease advances 3, 4. There is a growing interest in developing physical activity-based treatments such as tai chi and tango dancing that are designed to retrain balance and locomotion through repetitive movement5. However, there is limited evidence that these approaches are effective6. The focus of many physical therapy interventions is to promote rhythmic movement through cadenced repetition. We propose that cadenced repetition may contradict the pathophysiology of PD, which has been characterized as having greater-than-normal rhythmicity7. Normal gait involves a certain amount of variability, which is believed to optimize dynamic stability 8. So, we set out to describe the rhythmic characteristic of muscle activation signals in PD compared to a normative sample to provide deeper insight into the neuromuscular behaviour of PD gait disturbances.

Surface electromyography (SEMG) signals can provide a non-invasive representation of neural behaviour downstream from the central nervous system. Previous SEMG studies have identified abnormal motor control features of PD that have helped inform our understanding of the disease9, 10, 11. Ferrarin et al. (2007) identified the following abnormalities in EMG activity during PD gait12. There is prolonged activation of Rectus Femoris (RF) and Semimembranosus (SM) muscles throughout the gait cycle. Recruitment of the RF at heel contact and the SM in late swing are reduced. There is also reduced recruitment of the Gastrocnemius Medialis (GM) at push-off, and reduced recruitment of the Tibialis Anterior (TA) at heel contact. Some of these effects can be partially ameliorated with levodopa13. Premature activation of the TA and GM is seen prior to freezing of gait14. Miller et al. (1996) reported that variability in the gastrocnemius muscle was greater than normal in PD gait15. Taken together, these findings present a detailed but obscure picture of the pathological control of gait in PD.

There are a number of characteristic gait disturbances associated with PD, such as reduced arm swing, slow walking speed, short step length, and difficulty turning6, 16. PD gait is also characterized increased variability of walking speed, step length, step width, and stride duration17, 18. Fractal analyses have shown that the kinematics of PD gait are less ordered, and more complex and random-like compared to normal gait16, 19. All of these gait features negatively impact mobility and balance. They also contribute to postural instability and risk of falling, which can lead to further injury, disability and death4. In PD, the relative contribution to gait of supraspinal centres is altered20. It is not clear how or to what extent the gait disturbances are caused by impaired supraspinal input. The present study is one of the first attempts to analyze and interpret multiple muscle SEMG during PD gait.

The aim of this study was to describe rhythmic patterns of peripheral muscle activation in PD and identify differences from normative behaviour. The foundation of our analysis is the idea that spinal locomotor control consists of a set of modular burst generators21, 22. The idea that human gait is controlled by an efficient, minimal set of movement primitives, or muscle synergies, located in the spinal cord has been advanced for many years 23, 24. Muscle synergy analysis can be used by clinicians to draw inferences about neural structures underlying motor behaviours and guide rehabilitation decisions 25. There is some opposition, however, to the muscle synergy hypothesis in the scientific literature 26. Physiological interpretations based on the idea that reducible patterns in peripheral muscle activity reflect simplicity within the central nervous system should be made with caution.



Eight (n=8) individuals with idiopathic, mild to moderate PD (Hoehn-Yahr scale 1 to 3)27, and a convenience sample of ten (n=10) individuals with no history of neurological disorders (Able-bodied group) participated in this study. PD participants were recruited from an outpatient clinic of the Movement Disorders and Neurorehabilitation Center at the Methodist Neurological Institute in Houston, TX. All PD participants used Levodopa and performed all tasks while in the “ON” state. The study procedures were approved by the University of Houston’s Committee for Protection of Human Subjects, and all participants provided informed written consent. Descriptive statistics of the participants are provided in Table 1.

Table 1. Summary statistics of participants
Group N Age (years) Sex H&Y*
PD 8 71±12 6M/2F 2.2±0.7
AB 10 29±10 6M/4F

* Hoehn & Yahr scale19

Experimental Protocol

Each participant performed two walking trials over ground (OG), followed by two walking trials on a treadmill (TM). Between each trial, participants sat and rested for 5 minutes. The OG trials consisted of the participants walking along a straight 10m track on a hardwood floor in a large, open gymnasium. They started walking 2m before the track and continued walking for at least 2m after the track, such that the trial did not include step initiation or stopping. Participants were instructed to “walk as though you are going to mail a letter.” Their self-selected walking speed was calculated as 20m divided by the time it took to complete both OG trials (i.e., average speed of both trials). During the TM trials, the treadmill speed was set to the participant’s self-selected walking speed. Each TM trail was 60s in duration at a constant speed. An equal number of gait cycles were analysed under both conditions for each subject.


Participants’ lower extremities were instrumented with an 8-channel SEMG system (DataLOG MWX8, Biometrics Ltd, Ladysmith, VA, USA). Dry reusable surface electrodes (10mm diameter discs with 20mm between centres) were carefully placed over the muscle belly of the following muscles bilaterally: vastus lateralis (VL), long head of biceps femoris (BF), tibialis anterior (TA) and gastrocnemius lateralis (LG). The SENIAM recommendations for sensor location was used28. These particular muscles were selected as a representative set of the major actuators during gait29. The skin was cleaned and lightly abraded before the electrodes were attached with double-sided adhesive tape. SEMG signals were detected via differential electrodes, pre-amplified (1000 gain), filtered (bandpass: 20 – 450 Hz), and A/D converted at a sampling rate of 2000 Hz. The beginning and end of each gait cycle was determined by initial contact of the right foot, which was detected by the activation of a foot switch placed on the sole of the right shoe directly under the heel.

Data Analysis

After recording, SEMG signals were rectified and filtered using a zero-lag, low-pass Butterworth filter with a cut-off frequency of 10Hz, which is considered sufficient for noise removal without loss of signal30. Signals were then coded by fuzzy sets according to a classification procedure previously published31. Figure 1 illustrates the procedure. This method is designed to represent multiple muscle activation signals as a recurrent sequence of four basic burst patterns in the manner of a spinal locomotor circuit21. Four was shown to be the optimal number of bursts31. Pearson’s correlation coefficient, R, was determined using the empirical SEMG data versus the model estimation across all channels and gait cycles. The relative amount of SEMG signal that fits this model can be quantified using the statistic, R2, which is a measure of goodness of fit between the empirical data (Figure 2B) and the model-fitted signals (Figure 2D). R2 incorporated every individual gait cycle, and was calculated for each individual trial. We interpreted R2 as an index of rhythmicity, because it represents the proportion of the muscle activation signals that recur with observed regularity. Any stride-to-stride variability in muscle activation would not be represented by the model and would therefore decrease the index of rhythmicity.

Figure 1.Classification procedure. Processed SEMG signals (A) were separated into time-normalized gait cycles (B), and coded as four synergistic patterns of muscle activity during the gait cycle (C). Signals were reconstructed (D), and the amount of information retained was evaluated by the statistic, R2.
 Classification procedure. Processed SEMG signals (A) were separated into time-normalized gait cycles (B), and coded as four synergistic patterns of muscle activity during the gait cycle (C). Signals were reconstructed (D), and the amount of information retained was evaluated by the statistic, R2.

Figure 2.R2 for both groups, both walking conditions. Index of rhythmicity of muscle activation patterns for both groups and both conditions. Mean values for each group shown. Error bars represent standard deviations. Significant differences between groups and between walking conditions were observed
 R2 for both groups, both walking conditions. Index of rhythmicity of muscle activation patterns for both groups and both conditions. Mean values for each group shown. Error bars represent standard deviations. Significant differences between groups and between walking conditions were observed

A two-way analysis of variance (ANOVA) was performed with R2 as the dependent variable. We tested for main effects of one between-subject factor (PD versus Able-bodied), and one within-subject factor (OG versus TM). We also tested for an interaction effect of these two factors. Difference in self-selected walking speed between groups was tested using Student’s t-Test for independent means. A level of significance of p < 0.05 was used. All data processing and statistical analyses were performed using custom-written software in the MATLAB programming language (The Mathworks, Inc., Natick, MA, USA).


All participants successfully completed the protocol without difficulty. SEMG signal quality was good in all cases. The classification procedure yielded values of R2 from 0.582 to 0.866. The means and standard deviations for the different groups and walking conditions are illustrated in Figure 2. There was a significant group effect on the index of rhythmicity, R2 (p = .0458). The mean (± standard deviation) index of rhythmicity was 0.7754 ± 0.0574 for the PD patients, and 0.7138 ± 0.0683 for the able-bodied subjects. There was also a significant difference between OG walking and TM walking (p = .0097). The mean (± standard deviation) index of rhythmicity for both groups of participants was 0.7264 ± 0.0625 for OG walking, and 0.7559 ± 0.0757 for TM walking. There was no significant interaction between factors (p = .3671).

There was a significant difference between PD and the able-bodied group in terms of self-selected walking speed (p = .0024). The mean (± standard deviation) walking speed was 1.29 ± 0.18 m/s for the able-bodied subjects, and 0.87 ± 0.31 for the PD patients.


Healthy, mature motor skills are governed by neurological processes that allow for an optimal amount of movement variability32. Systems that produce more than optimal variability are noisy and unstable, while systems that produce less than optimal variability tend to be rigid and unfit to cope with perturbations. Pathological conditions can result in either case.

The main finding of this study is that the rhythmicity of muscle activation signals in the lower extremities of elderly individuals with PD is higher than in young, able-bodied individuals. In other words, the muscle activation patterns in PD exhibit greater regularity and are more consistent with the four-burst model of spinal locomotor control. It also means that there was less stride-to-stride variability in terms of SEMG in PD patients than in able-bodied individuals. In normal gait, the majority of the SEMG signals (70% OG, 72% TM) were explained by the four-burst CPG model. The parts of the signal that were not explained by the model would be made up of non-rhythmic elements such as, anticipatory postural adjustments, feedback response to irregular perturbations, and aberrant neurological commands (i.e., noise). In PD gait, a significantly greater portion of the SEMG signals is represented by the CPG model (76% OG, 80% TM). This is a rather unexpected result. We presumed that PD gait would involve more aberrant neurological commands than normal, based on reports that there is higher variability in PD gait 15, 16, 33. However, the overall amount of non-rhythmic elements in SEMG was reduced in PD gait, suggesting that there are fewer stride-to-stride adjustments and responses to errors.

The idea that human locomotion is driven by oscillating neural circuits located in the spinal cord has been advanced for decades34. These circuits, known as the Central Pattern Generator (CPG), provide rhythmic bursts of muscle activation signals that form the basis of locomotor control35, 36, 37. Rhythmic patterns can be identified using a statistical model of CPG31. By this approach, we are able to estimate the relative contribution of the CPG to muscle activation signals. Furthermore, we can estimate how much of the variability in gait is due to sources other than the spinal CPG circuits (e.g., feedforward adjustments from supraspinal centers).

If confirmed by a larger study, these findings may lead to significant changes in current clinical approaches to gait retraining of patients with PD. The traditional physical therapy approach of retraining locomotion through rhythmic activities such as treadmill training, tandem bicycling, and OG walking may be inconsistent with the fact that rhythmic muscle activation patterns are already more pronounced in PD5, 6. These findings may support the development of new research protocols for assessing the role of conventional therapies, such as dopamine replacement therapy and DBS, on neurological control of gait.

One of our secondary findings was that there is a significant difference between OG walking and TM walking in both groups. Muscle activation signals during TM were more rhythmic than during OG walking. This result was expected. TM walking involves a more stable, controlled environment with fewer irregularities to perturb the body. Therefore, there is less need for anticipatory adjustments from the supraspinal control centres, and the spinal CPG output will be more represented in the muscle activation signals.

The variability of PD gait has been reported broadly in the scientific literature. Most analyses have focused on the variability of kinematic features of gait, i.e., stride length, stride interval, gait speed, etc.17, 18. These studies have all concluded that the variability of movement in PD gait is greater than in normal gait. However, none of them address the cause of the movement variability. Our findings suggest that there is reduced variability in the control signals that are sent to the skeletal muscles. This does not contradict the previous observations of increased movement variability.

It should be noted that comparisons between the PD group and the young, able-bodied controls could be confounded by age. It is not possible to determine from our data whether between groups differences are due to age, disease or something else. The Able-bodied data is provided as a reference of normative behavior.

The present study is one of the few analyses that deal with variability of multiple muscle activation signals using SEMG. Some studies have focused on individual SEMG signals. Previously, it was reported that the variability of SEMG in the GM muscle is greater in PD than in able-bodied individuals12. However, when we looked at multiple muscle SEMG signals, we found that they were much more representative of basic CPG function. There is no inconsistency between these two different findings. Our methods represent a broader paradigm which is more consistent with the complexity of CNS control of gait.


Individuals with PD appear to walk with muscle activation patterns that are more rhythmic than those seen in young, able-bodied individuals. Specifically, their muscle activation patterns exhibit more stride-to-stride consistency and are easier to describe with a basic burst model of recurring muscle activation. According to the optimal variability concept32, this behaviour is characteristic of a rigid control system, which is less able to cope with perturbations than a more flexible system. One possible explanation for this is that PD gait is more heavily influenced by spinal locomotor control with diminished control elements from supraspinal centres.


The authors acknowledge the financial support of the University of Houston department of Health and Human Performance.


  1. 1.LML de Lau, Breteler MMB.. , Epidemiology of Parkinson’s disease,” Lancet Neurol,Jan.2006 5, 525-35.
  1. 2.Fritsch T, K a Smyth, M S Wallendal, Hyde T, Leo G et al.Parkinson disease: research update and clinical management.,”. , South. Med. J,Dec.2012 105(12), 650-6.
  1. 3.Herman T, Giladi N, J M Hausdorff.Treadmill training for the treatment of gait disturbances in people with Parkinson’s disease: a mini-review.,”. , J. Neural Transm.Mar2009 116(3), 307-18.
  1. 4.Nutt J G, FBHorak, Bloem B R.Milestones in gait, balance. , and falling.,” Mov. Disord,May2011 26(6), 1166-74.
  1. 5.Jahn K, Zwergal A.Imaging supraspinal locomotor control in balance disorders.,”. , Restor. Neurol. Neurosci,Jan2010 28(1), 105-14.
  1. 6.C L Tomlinson, Patel S, Meek C, C P Herd, C E et al.Physiotherapy intervention in Parkinson’s disease: systematic review and meta-analysis,” BMJ,6Aug2012. 345(1), 5004-5004.
  1. 7.Wu Y, Krishnan S, Member S. (2010) Statistical Analysis of Gait Rhythm. in Patients With Parkinson’s Disease,” Rehabilitation 18(2), 150-158.
  1. 8.Stergiou N, Decker L M.Human movement variability, nonlinear dynamics, and pathology: Is there a connection?,”. , Hum. Mov. Sci,Jul2011 30(5), 869-888.
  1. 9.PHJ a Nieuwenhuijzen, Horstink M W, Bloem B R, Duysens J.Startle responses in Parkinson patients during human gait.,”. Exp. brain Res,May2006 171(2), 215-24.
  1. 10.Dietz V, Colombo G.Influence of body load on the gait pattern. in Parkinson’s disease.,” Mov. Disord,Mar1998 13(2), 255-61.
  1. 11.Frigo C, Crenna P.Multichannel SEMG in clinical gait analysis: a review and state-of-the-art.,”. , Clin. Biomech,Mar2009 24(3), 236-45.
  1. 12.Ferrarin M, Carpinella I, Rabuffetti M, Rizzone M, Lopiano L. (2007) Unilateral and Bilateral Subthalamic Nucleus Stimulation in Parkinson’s Disease:. , Effects on EMG Signals of Lower Limb Muscles During Walking,” IEEE Trans Neural Syst Rehabil Eng 15(2), 182-189.
  1. 13.Caliandro P, Ferrarin M, Cioni M, A R Bentivoglio, Minciotti I et al.Levodopa effect on electromyographic activation patterns of tibialis anterior muscle during walking. in Parkinson’s disease.,” Gait Posture,Mar2011 33(3), 436-41.
  1. 14.Nieuwboer A, Dom R, W De, Desloovere K, Janssens L et al.Electromyographic profiles of gait prior to onset of freezing episodes. in patients with Parkinson’s disease.,” Brain,Jul2004 127, 1650-60.
  1. 15.Miller R A, Thaut M H, McIntosh G C, Rice R R. (1996) Components of EMG symmetry and variability in parkinsonian and healthy elderly gait,”. , Electroencephalogr Clin Neurophysiol 101(1), 1-7.
  1. 16.J M Hausdorff.Gait dynamics in Parkinson’s disease: common and distinct behavior among stride length, gait variability, and fractal-like scaling.,”. , Chaos,Jun2009 19(2), 026113.
  1. 17.Wu Y, Krishnan S. (2010) Statistical Analysis of Gait Rhythm in Patients With Parkinson’s Disease,”. , IEEE Trans Neural Syst Rehabil Eng 18(2), 150-158.
  1. 18.Frenkel-Toledo S, Giladi N, Peretz C, Herman T, Gruendlinger L et al. (2005) Effect of gait speed on gait rhythmicity in Parkinson’s disease: variability of stride time and swing time respond differently,”. , J Neuroeng Rehabil 2, 23.
  1. 19.Sekine M, Tamura T, Akay M, Fujimoto T, Togawa T et al.Discrimination of walking patterns using wavelet-based fractal analysis.,”. , IEEE Trans Neural Syst Rehabil Eng,Sep2002 10(3), 188-96.
  1. 20.Takakusaki K, Tomita N, Yano M.Substrates for normal gait and pathophysiology of gait disturbances with respect to the basal ganglia dysfunction.,”. , J. Neurol,Aug2008 255, 19-29.
  1. 21.S M Danner, U S Hofstoetter, Freundl B, Binder H, Mayr W et al. (2015) Human spinal locomotor control is based on flexibly organized burst generators,”. , Brain 138, 577-588.
  1. 22.Bizzi E, D’Avella A, Saltiel P, Tresch M. (2002) Modular organization of spinal motor systems,”. , Neuroscientist 8(5), 437-442.
  1. 23.F De Groote, Jonkers I, Duysens J.Task constraints and minimization of muscle effort result in a small number of muscle synergies during gait.,”. , Front. Comput. Neurosci,September2014 8, 115.
  1. 24.K M Steele, Rozumalski A, M H Schwartz. (2015) Muscle synergies and complexity of neuromuscular control during gait in cerebral palsy,”. , Dev. Med. Child Neurol 1-7.
  1. 25.Safavynia S A, Torres-Oviedo G, Ting L H. (2011) Muscle Synergies: Implication for Clinical Evaluation and. , Rehabilitation of Movement,” NIH-PA Author Manuscr 17(1), 16-24.
  1. 26.M C Tresch, Jarc A. (2009) The case for and against muscle synergies,”. , Curr. Opin. Neurobiol 19(6), 601-607.
  1. 27.Hoehn M M, Yahr M D. (1967) Parkinsonism: onset, progression. , and mortality,” Neurology 17(5), 427-442.
  1. 28.. SENIAM”[Online].Available:
  1. 29.B H Jansen, V H Miller, D C Mavrofrides, Jansen C W Stegink.Multidimensional EMG-based assessment of walking dynamics.,”. , IEEE Trans Neural Syst Rehabil Eng,Sep2003 11(3), 294-300.
  1. 30.Shiavi R, Frigo C, Pedotti a.Electromyographic signals during gait: criteria for envelope filtering and number of strides.,”. , Med. Biol. Eng. Comput,Mar1998 36(2), 171-8.
  1. 31.Thrasher T A, Ward J S, Fisher S.Classification of rhythmic locomotor patterns in electromyographic signals using fuzzy sets.,”. , J. Neuroeng. Rehabil,Dec2011 8(1), 65.
  1. 32.Stergiou N, Harbourne R, J.Cavanaugh.Sep2006 “Optimal movement variability: a new theoretical perspective for neurologic physical therapy.,”. , J. Neurol. Phys. Ther 30(3), 120-9.
  1. 33.Nieuwboer A, Dom R, W De, Desloovere K, Janssens L et al. (2004) Electromyographic profiles of gait prior to onset of freezing episodes in patients with Parkinson’s disease,”. 127, 1650-1660.
  1. 34.MM Pinter Aug1999.Gait after spinal cord injury and the central pattern generator for locomotion,”. , Spinal Cord 37(8), 531-537.
  1. 35.Y P Ivanenko, Cappellini G, Dominici N, R E Poppele, Lacquaniti F. (2005) Coordination of locomotion with voluntary movements in humans,”. , J Neurosci 25(31), 7238-7253.
  1. 36.Y P Ivanenko, R E Poppele, Lacquaniti F.Distributed neural networks for controlling human locomotion: lessons from normal and SCI subjects.,” Brain Res. , Bull,Jan.2009 78(1), 13-21.
  1. 37.B W Gallarda, T O Sharpee, S L Pfaff, W A Alaynick. (2010) Defining rhythmic locomotor burst patterns using a continuous wavelet transform,”. , Ann N Y Acad Sci 1198, 133-139.