Model for prompt and effective classification of motion recovery after stroke considering muscle strength and coordination factors
Muscle synergies are now widely discussed as a method for evaluating the existence of redundant neural networks that can be activated to enhance stroke rehabilitation. However, this approach was initially conceived to study muscle coordination during learned motions in healthy individuals. After brain damage, there are several neural adaptations that contribute to the recovery of motor strength, with muscle coordination being one of them. In this study, a model is proposed that assesses motion based on surface electromyography (sEMG) according to two main factors closely related to the neural adaptations underlying motor recovery: (1) the correct coordination of the muscles involved in a particular motion and (2) the ability to tune the effective strength of each muscle through muscle fiber contractions. These two factors are hypothesized to be affected differently by brain damage. Therefore, their independent evaluation will play an important role in understanding the origin of stroke-related motor impairments.
The model proposed was validated by analyzing sEMG data from 18 stroke patients with different paralysis levels and 30 healthy subjects. While the factors necessary to describe motion were stable across heathy subjects, there was an increasing disassociation for stroke patients with severe motor impairment.
The clear dissociation between the coordination of muscles and the tuning of their strength demonstrates the importance of evaluating these factors in order to choose appropriate rehabilitation therapies. The model described in this research provides an efficient approach to promptly evaluate these factors through the use of two intuitive indexes.
KeywordsMotion performance Stroke recovery Electromyography Muscle synergies Muscle mirror symmetry Muscle effective strength Motor strength gain
Effective strength balance
Muscle coordination similarity
Root mean squares
Stroke impairment assessment set
As a consequence of the population aging around the world, stroke has become a widespread concern . Depending on the brain area and size of the injury, the consequences of motor impairments vary significantly . It has been widely shown that in order to enhance stroke rehabilitation with respect to effective motor recovery, it is necessary to start therapy shortly after the cerebrovascular accident [3, 4, 5]. Therefore, the use of inefficient treatments during early stages of rehabilitation, might result in an insufficient motor recovery. More than 60% of stroke survivors have remaining motor paralysis, resulting in serious social cost and affecting their quality of life for the rest of their lives [6, 7].
In recent years, the concept of muscle synergies has been widely discussed to clarify the biological basis for activating the redundant musculoskeletal system of stroke survivors [8, 9, 10]. According to this model, during motor learning, the plastic nature of neurons creates modules or neural networks (called synergies) that are specialized for different tasks [10, 11]. Synergies control the contraction of a set of muscles using a low-dimensional set of control commands originating from the brain [12, 13, 14, 15]. It has been proposed that use of the remaining neural pathways could potentiate post-stroke recovery. A study on stroke patients with affected frontal motor cortical areas shows similar synergies in both arms (i.e., paretic and non-paretic) irrespective of their motion performance , suggesting a synergetic behavior independent of task constraints. In addition, another study  reveals the natural emergence of new muscle synergies during the learning process associated to the control of a myoelectric interface. However, these findings conflicts with other studies that show abnormal synergies appearing for the paretic arm of stroke patients [9, 18, 19]. This phenomenon was further evaluated by  in 31 stroke survivors. Paretic and non-paretic upper limb synergies were compared, showing three different behaviors on the non-paretic side: preservation, merging, and fractionation of synergies. Each behavior was affected differently by the level of motor impairment and poststroke duration. These results predict a wide range of stroke conditions from the perspective of muscle synergies, highlighting the importance of individual patient-by-patient evaluations. In order to clarify the origin of these results, it is necessary to understand the reasons behind the development of the muscle synergy approach and contrast them with the main concerns in the field of stroke rehabilitation.
Learned motions are perceived by healthy individuals as simple and easy, even though they require the complex coordination of muscles. The concept of muscle synergies was originally developed to explain the neural processes that provide the brain with tools to reduce the kinesiological complexity behind this coordination. However, after a stroke, the failure in motor control occurs not in the spinal cord but in the brain. Although studying the appearance of new muscle synergies after brain damage is necessary, current knowledge on this topic is not enough to choose an appropriate rehabilitation treatment. Neural recovery starts at the brain level, and depending on the severity of the injury will trigger changes at the spinal and muscular levels . In this scenario, motor recovery relies not only on the neural process of motor coordination but also on all the physiological processes underlying motor strength gaining . It is currently agreed that gains in voluntarily motor strength depend on two main factors: neural adaptations and muscle hypertrophy . Brain damage alone does not affect muscle fiber condition; therefore, the loss of motor strength directly after a stroke is related to malfunctions of neural adaptations. The current literature enumerates these issues as malfuncitons of potentiation of neural connectivity, motor unit synchronization, muscle coordination, and learning . The study of these neural adaptations during motion recovery is a challenging task due to the complexity of the neural system and current technological limitations on their measurement. However, it is possible to infer their effects on surface electromyography (sEMG) signals by evaluating muscle activation according to the following two factors. (1) The neural adaptations in charge of muscle coordination and learning affect the distribution of electrical power among all muscles contributing to the motion. (2) Structural and synaptic neural connectivity and the processes underlying the recruitment of motor units affect the ability to tune the effective strength of each muscle contraction. Assuming the symmetrical properties of the healthy human body, a in this study a model is proposed for evaluating these two factors during periodic symmetrical motions through the analysis of sEMG signals.
During motion in healthy individuals, the mechanisms of muscle coordination and the tuning of effective muscle strength compensate for each other, thereby creating stable muscle synergies. In this research, it is hypothesized that after brain damage, these two factors are affected differently (i.e., disassociated), thus generating a wider range of motor impairments. As these factors are directly related to the neural adaptations underlying motor recovery, their measurement will facilitate the prompt identification of the neural origins of the motor impairment, and therefore the selection of an effective rehabilitation therapy.
Materials and methods
Motions and experimental procedure
The experimental trial consisted of 20 s of motion followed by 20 s of rest. The experimental session was composed of three 20-s elbow flexion trials and three 20-s dual steering trials. These trials were presented in a randomized order to reduce possible biases in subjects’ motor behavior owing to constant repetition of the same task.
Muscle activity was recorded using 18 wireless sEMG sensors (BTS FREEEMG; BTS Bioengineering Corp., Milan, Italy) symmetrically located on the following nine upper limb muscles (on both the left and right sides of the body): brachioradialis, pronator teres, biceps, triceps, anterior deltoid, posterior deltoid, pectoralis, infraspinatus, and elector spinae. These muscles were defined according to the guidelines of the Surface Electromyography for the Non-Invasive Assessment of Muscle Project . Acquired data were digitalized using a sampling frequency of 1000 Hz and stored for subsequent processing.
Basic sEMG processing
Raw sEMG data were high-pass filtered at 20 Hz to remove possible motion artifacts, rectified, and low-pass filtered at 32 Hz to preserve the key frequencies associated with muscle contractions. In both filtering stages, a fifth-order Butterworth filter was applied . In addition, all channel amplitudes were standardized according to the median values recorded during the experimental session over the whole set of channels .
Approach applied during symmetric motions
After sEMG processing, the power per second of each muscle was quantized as the RMS of the rectified signal. In addition, the indexes defined in the motion model described in the previous section were computed. Coefficients ai and bi were extracted to show the benefits of their evaluation in assessing the importance of each muscle depending on the motion. Sets of ai and bi were represented as 18 values (associated with the nine muscles measured on each side of the body). In addition, the values of ESB and MCS were computed over the three groups of subjects to assess how different severity levels are affected by the two groups of neural adaptations that participate in motor performance.
Muscle activity prints
The healthy young and elder muscle prints show how the important muscles involved in voluntarily motor control differ between the movements performed. During elbow flexion (Fig 3a), biceps and anterior deltoid muscles are dominant but strongly supported by brachioradialis, pronator teres, triceps, and pectoralis muscles. For the steering motion (Fig 3b), anterior deltoid and pectoralis muscles are the main carriers of the movement, while the remaining muscles contribute less power. This is expected as the dual steering system provides partial gravity support, which compensates for the power of supporting muscles used in the same task. Moreover, the rarity of outliers (red dots) among the healthy subjects indicates less variability on the sEMG data analyzed and therefore suggests the existence of stable muscle behaviors among healthy subjects associated with the specific motions evaluated. For stroke patients, even similar muscle trends can be observed on average (and for both motions), but the expected muscle prints are more difficult to distinguish. In addition, the number of outliers was drastically higher among patients, which is expected for the stroke population. Depending how the brain injury affects the neural processes controlling motion, each patient developed different control strategies manifested as muscle asymmetries. For example, in the case of elbow flexion, outliers demonstrate the tendency of some patients to use their posterior deltoid muscles to compensate for the lack of movement of the main muscles (i.e., anterior deltoid and biceps).
Effective strength balance and muscle coordination similarity
SIAS versus MCS and ESB
It was initially argued that the mere evaluation of muscle coordination in patients recovering from strokes is insufficient to elucidate all neural processes involved in motor recovery. Even though the proper coordination of muscles is an important factor that contributes to the performance of an efficient motion, the ability to properly tune the effective strength of muscle contractions also plays an important role. By exploiting the symmetry of the human body, the model described in this work proposes two indexes directly related to the factors involved in motor recovery. The computation of these indexes for subjects ranging from healthy individuals to severe stroke patients provided interesting muscle performance results in terms of the factors involved in motor recovery.
For the healthy young and elder subjects, there were clear similarities between body sides both in muscle coordination (MCS) and muscle effective strength (ESB) (Fig. 4a-c). Also, the statistical comparison performed between these two groups on Fig. 5 shows no significant differences, suggesting the absence of age dependence on these factors during simple symmetrical motions. The ability to efficiently coordinate muscles and tune their contraction strength creates a perfect scenario for the study of muscle synergies. Healthy subjects are able to maintain stable synergies throughout a range of variation within a task by increasing or decreasing motor unit recruitment, thereby tuning the strength of individual muscles.
However, in the case of stroke survivors, the ability to coordinate muscles and tune their effective contraction strength are affected differently depending on the area and size of the lesion. This is clearly represented by MCS and ESB indexes in Fig. 4b–d. Even within patients classified with the same paralysis level, their performance regarding these two factors hugely differs. The comparative example highlighted in Fig. 4d shows how a patient with almost healthy muscle coordination (dashed-circle 1) is classified as SIAS 1 owing to their inability to balance strength among sides of the body. At the same time, another patient, also classified as SIAS 1, shows no problems balancing the strength among sides but completely lacks the ability to coordinate their muscles properly (dashed-circle 2). Moreover, by evaluating the standard deviations shown in Fig. 5, it can be seen that both MCS and EBS present increasing dispersion for patients with lower SIAS levels. This suggests that muscle coordination and muscle strength tuning are less coupled in severe patients, contributing to an explanation of the appearance of a wide range of synergistic abnormalities when evaluating them.
Accordingly, the application of models that explain motor control based on neural processes in charge of motor coordination is incomplete in studying motor recovery. In this scenario, other neural processes often related to recovery of motion strength also play an important role. The identification of the neurological origins of post-stroke motor impairment is a relevant factor for the assignment of effective rehabilitation therapies. In that regard, as supported by the data analyzed, it appears that the model described and the indexes inferred from it have the potential to distinguish among the different factors affecting motor recovery based on the analysis of sEMG signals. In addition, the reduced setup and simplicity of calculating these indexes make them easy tools for standardizing rehabilitation therapies.
This research focused on the development of a model for extracting information from sEMG signals to quantify the neural mechanisms behind motor recovery. Given the physiological properties of sEMG signals and current knowledge about the neural processes in charge of recovering motor strength [22, 23, 23], motion performance was defined as the combination of two main factors: (1) coordination of muscles (representing the synergetic behaviors observed in healthy subjects) and (2) tuning of strength through individual muscle contractions (representing the neural processes underlying recruiting motor units and strengthening neural synaptic and structural connections). The model-inferred indexes MCS and ESB, which represent these factors, were validated using sEMG data recorded from three different groups of subjects during two different motions. Even though these factors seemed closely related during healthy subject motion, results obtained from stroke patients suggest that brain damage contributes to decoupling both factors, thereby creating a range of motor impairments that are commensurate with stroke severity. Current standard stroke rehabilitation therapies are focussed on the improvement of muscle coordination [42, 43], however our results show that some patients with poor motor performance show good muscle coordination (MCS). This fact suggest that a direct use of the metrics developed could be used to choose between a muscle coordination or muscle strength oriented therapy.
Finally, it is important to note that the present method is based on a comparison between paretic and non-paretic body sides. Even though this is a commonly accepted methodology in stroke research, it assumes a healthy condition of the non-paretic side. To avoid the misinterpretation of results, researchers using these parameters should confirm the condition of the non-paretic area used as the empirical expectation. However, the current work is not intended to study the rehabilitation benefits of the specific motions evaluated. The recorded sEMG data were used only to validate the model proposed and show its potential as a tool to evaluate motor recovery. The model consistency should be also further analyzed by including a wider range of tasks to those evaluated on this work. It should be also mention that the methodology proposed do not intend to provide ultimate solutions to all current challenges in the field of motor rehabilitation. For instance, many daily tasks are unimanual or require independent actions on each limb. The system proposed, as happens also with many other rehabilitation techniques, cannot be used on a daily environment. However, it allows the use of the non-paretic areas of the body, within a controlled clinical environment, in order to potentiate the recovery of the paretic areas by exploiting the symmetrical structure of human body. In fact, the recovery of specific motions on the paretic limb using feedback from the non-paretic side is a common technique on rehabilitation . Nevertheless, the use of our model for unimanual task could be discussed for future studies focused on increasing patients’ motion dexterity.
This work is the beginning of a wider research program that is intended to increase the current understanding of motor recovery in order to improve rehabilitation therapies. Accordingly, the next research stages include the following objectives: (1) development of a system that provides real-time feedback about ESB and MCS to patients during rehabilitation; (2) integration of ESB and MCS with the study of muscle synergy development during rehabilitation to better understand the neural processes following recovery; and (3) development of a rehabilitation protocol adaptable to specific patient conditions and neural recovery strategies based on the results obtained in the this and future studies.
Toyota Motor Corporation designed the dual steering system.
ACG developed the mathematical concept of the model described in the paper. ACG, HY, MI, FA, SO, and SS developed the experimental protocol and conducted the experiments with healthy subjects. KO, MT, and IK provided and scheduled the experimentations with stroke patients and healthy elder subjects. Experiments with both groups were performed with the collaboration of ACG, HY, MI, FA, SO, and SS. ACG analyzed and processed the data. ACG and SS drafted the structure of the manuscript, while the final version was obtained after collaboration and review by all authors. All authors read and approved the final manuscript.
Toyota Motor Corporation provided the founding to hire the human resources in charge of the current research in collaboration with RIKEN.
Ethics approval and consent to participate
All subjects were previously informed about the experimental procedure and provided written informed consent, in accordance with the Declaration of Helsinki. The experimental procedure was approved by the ethics committees of the National Center for Geriatrics and Gerontology and RIKEN.
Consent for publication
The subject used as model in Fig. 1 provided written consent for the publication of the images shown.
The authors declare that they have no competing interests.
- 25.Johnson M, Paranjape R, Strachota E, Tchekanov G, McGuire J. Quantifying learned non-use after stroke using unilateral and bilateral steering tasks. In: 2011 IEEE International Conference on Rehabilitation Robotics. IEEE: 2011. p. 1–7. https://doi.org/10.1109/icorr.2011.5975457.
- 27.Malabet HG, Robles RA, Reed KB. Symmetric motions for bimanual rehabilitation. In: 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE: 2010. p. 5133–8. https://doi.org/10.1109/iros.2010.5651239.
- 29.Whitall J, Waller SM, Sorkin JD, Forrester LW, Macko RF, Hanley DF, Goldberg AP, Luft A. Bilateral and unilateral arm training improve motor function through differing neuroplastic mechanisms: a single-blinded randomized controlled trial. Neurorehabil Neural Repair. 2011; 25(2):118–29.PubMedCrossRefGoogle Scholar
- 31.Hermens HJ, Freriks B, Merletti R, Stegeman D, Blok J, Rau G, Disselhorst-Klug C, Hägg G. European recommendations for surface electromyography. Roessingh Res Dev. 1999; 8(2):13–54.Google Scholar
- 41.Abdi H. Holm’s sequential bonferroni procedure. Encycl Res Des. 2010; 1(8):1–8.Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.