skip to main content

Title: Abstracts From the 2018 Annual Meeting
Muscle weakness and loss of independent joint control are the 2 most common neuromotor impairments after stroke. While there are a number of approaches to improve poststroke muscle weakness, there are currently no rehabilitation strategies that directly target a patient’s inability to match and independently activate the normal patterned muscle coordination strategies, or “muscle synergies.” Our goal is to develop an EMG-based controller for retraining healthy muscle synergies in patients with stroke-related disabilities. The controller can be integrated into rehabilitation robots for their ability to structure the robot’s force output based on input EMG activity. However, developing such a controller would require a clear understanding of the relationship between the applied force from a rehabilitation robot and the resulting changes to a patient’s muscle synergies. Therefore, this study was performed to quantify how the muscle synergies of horizontal planar-reaching are affected by direction of an applied force at the end-effector (ie, hand). A 2 DOF, 10 muscle model was developed in MATLAB using parameters obtained from the OpenSim (version 3.3) open source software system. Simulation experiments were then performed in MATLAB to investigate the relationship between the applied force and the resulting muscle synergies. The simulated event was composed of several trials of the same righthanded, planar, multidirectional reaching task from 0° (to the right) to 360°. Each trial applied a different steering force direction at the subject’s hand, varying from −45° to 45° relative to the reaching direction. The simulation trials were also validated by evaluating the EMG patterns of a healthy subject when performing the same reaching task with varying steering force directions. For the 0° steering force trials, the muscle synergies and their activation timings were extracted using nonnegative matrix factorization (NMF). For all other trials, the synergy matrix was fixed and the activation timings were extracted from the product of the EMG of that trial and the pseudo-inverse of the synergy matrix from the 0° steering force trial. By fixing the synergy matrix in the trials with steering forces, we can directly track activation changes of a certain synergy as steering force is varied. For both simulation and experimental trials, circular statistics revealed a linear relationship between changes in steering force direction and principal direction of synergy activation. These results suggest that the activation of a synergy can be controlled directly by the direction of an applied steering force. This has relevant implications in synergy-based controller design because a computer can easily manipulate a patient’s muscle synergies and track the changes while avoiding the computational expense of NMF. In addition, similar analysis could be used to extract the relationship between applied forces and changes in synergies for other types of motion.  more » « less
Award ID(s):
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Neurorehabilitation and Neural Repair
Page Range / eLocation ID:
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Abstract Handedness has been associated with behavioral asymmetries between limbs that suggest specialized function of dominant and non-dominant hand. Whether patterns of muscle co-activation, representing muscle synergies, also differ between the limbs remains an open question. Previous investigations of proximal upper limb muscle synergies have reported little evidence of limb asymmetry; however, whether the same is true of the distal upper limb and hand remains unknown. This study compared forearm and hand muscle synergies between the dominant and non-dominant limb of left-handed and right-handed participants. Participants formed their hands into the postures of the American Sign Language (ASL) alphabet, while EMG was recorded from hand and forearm muscles. Muscle synergies were extracted for each limb individually by applying non-negative-matrix-factorization (NMF). Extracted synergies were compared between limbs for each individual, and between individuals to assess within and across participant differences. Results indicate no difference between the limbs for individuals, but differences in limb synergies at the population level. Left limb synergies were found to be more similar than right limb synergies across left- and right-handed individuals. Synergies of the left hand of left dominant individuals were found to have greater population level similarity than the other limbs tested. Results are interpreted with respect to known differences in the neuroanatomy and neurophysiology of proximal and distal upper limb motor control. Implications for skill training in sports requiring dexterous control of the hand are discussed. 
    more » « less
  2. Abstract Background Few studies have systematically investigated robust controllers for lower limb rehabilitation exoskeletons (LLREs) that can safely and effectively assist users with a variety of neuromuscular disorders to walk with full autonomy. One of the key challenges for developing such a robust controller is to handle different degrees of uncertain human-exoskeleton interaction forces from the patients. Consequently, conventional walking controllers either are patient-condition specific or involve tuning of many control parameters, which could behave unreliably and even fail to maintain balance. Methods We present a novel, deep neural network, reinforcement learning-based robust controller for a LLRE based on a decoupled offline human-exoskeleton simulation training with three independent networks, which aims to provide reliable walking assistance against various and uncertain human-exoskeleton interaction forces. The exoskeleton controller is driven by a neural network control policy that acts on a stream of the LLRE’s proprioceptive signals, including joint kinematic states, and subsequently predicts real-time position control targets for the actuated joints. To handle uncertain human interaction forces, the control policy is trained intentionally with an integrated human musculoskeletal model and realistic human-exoskeleton interaction forces. Two other neural networks are connected with the control policy network to predict the interaction forces and muscle coordination. To further increase the robustness of the control policy to different human conditions, we employ domain randomization during training that includes not only randomization of exoskeleton dynamics properties but, more importantly, randomization of human muscle strength to simulate the variability of the patient’s disability. Through this decoupled deep reinforcement learning framework, the trained controller of LLREs is able to provide reliable walking assistance to patients with different degrees of neuromuscular disorders without any control parameter tuning. Results and conclusion A universal, RL-based walking controller is trained and virtually tested on a LLRE system to verify its effectiveness and robustness in assisting users with different disabilities such as passive muscles (quadriplegic), muscle weakness, or hemiplegic conditions without any control parameter tuning. Analysis of the RMSE for joint tracking, CoP-based stability, and gait symmetry shows the effectiveness of the controller. An ablation study also demonstrates the strong robustness of the control policy under large exoskeleton dynamic property ranges and various human-exoskeleton interaction forces. The decoupled network structure allows us to isolate the LLRE control policy network for testing and sim-to-real transfer since it uses only proprioception information of the LLRE (joint sensory state) as the input. Furthermore, the controller is shown to be able to handle different patient conditions without the need for patient-specific control parameter tuning. 
    more » « less
  3. Converging evidence in human and animal models suggests that exogenous stimulation of the motor cortex (M1) elicits responses in the hand with similar modular structure to that found during voluntary grasping movements. The aim of this study was to establish the extent to which modularity in muscle responses to transcranial magnetic stimulation (TMS) to M1 resembles modularity in muscle activation during voluntary hand movements involving finger fractionation. EMG was recorded from eight hand-forearm muscles in nine healthy individuals. Modularity was defined using non-negative matrix factorization to identify low rank approximations (spatial muscle synergies) of the complex activation patterns of EMG data recorded during high density TMS mapping of M1 and voluntary formation of gestures in the American Sign Language alphabet. Analysis of synergies as a set, and individually, revealed greater than chance similarity between those derived from TMS and those derived from voluntary movement. Both datasets included synergies dominated by single intrinsic hand muscles presumably to meet the demand for highly fractionated finger movement. These results suggest a cortical role in combining corticospinal connectivity to individual intrinsic hand muscles with modular mulit-muscle activation via synergies. 
    more » « less
  4. Abstract Introduction Split-belt treadmill training has been used to assist with gait rehabilitation following stroke. This method modifies a patient’s step length asymmetry by adjusting left and right tread speeds individually during training. However, current split-belt training approaches pay little attention to the individuality of patients by applying set tread speed ratios (e.g., 2:1 or 3:1). This generalization results in unpredictable step length adjustments between the legs. To customize the training, this study explores the capabilities of a live feedback system that modulates split-belt tread speeds based on real-time step length asymmetry. Materials and methods Fourteen healthy individuals participated in two 1.5-h gait training sessions scheduled 1 week apart. They were asked to walk on the Computer Assisted Rehabilitation Environment (CAREN) split-belt treadmill system with a boot on one foot to impose asymmetrical gait patterns. Each training session consisted of a 3-min baseline, 10-min baseline with boot, 10-min feedback with boot (6% asymmetry exaggeration in the first session and personalized in the second), 5-min post feedback with boot, and 3-min post feedback without boot. A proportional-integral (PI) controller was used to maintain a specified step-length asymmetry by changing the tread speed ratios during the 10-min feedback period. After the first session, a linear model between baseline asymmetry exaggeration and post-intervention asymmetry improvement was utilized to develop a relationship between target exaggeration and target post-intervention asymmetry. In the second session, this model predicted a necessary target asymmetry exaggeration to replace the original 6%. This prediction was intended to result in a highly symmetric post-intervention step length. Results and discussion Eleven out of 14 participants (78.6%) developed a successful relationship between asymmetry exaggeration and decreased asymmetry in the post-intervention period of the first session. Seven out of the 11 participants (63.6%) in this successful correlation group had second session post-intervention asymmetries of < 3.5%. Conclusions The use of a PI controller to modulate split-belt tread speeds demonstrated itself to be a viable method for individualizing split-belt treadmill training. 
    more » « less
  5. Abstract Background

    Ankle-targeting resistance training for improving plantarflexion function during walking increases rehabilitation intensity, an important factor for motor recovery after stroke. However, understanding of the effects of resisting plantarflexion during stance on joint kinetics and muscle activity—key outcomes in evaluating its potential value in rehabilitation—remains limited. This initial study uses a unilateral exosuit that resists plantarflexion during mid-late stance in unimpaired individuals to test the hypotheses that when plantarflexion is resisted, individuals would (1) increase plantarflexor ankle torque and muscle activity locally at the resisted ipsilateral ankle, but (2) at higher forces, exhibit a generalized response that also uses the unresisted joints and limb. Further, we expected (3) short-term retention into gait immediately after removal of resistance.


    Ten healthy young adults walked at 1.25 m s−1for four 10-min discrete bouts, each comprising baseline, exposure to active exosuit-applied resistance, and post-active sections. In each bout, a different force magnitude was applied based on individual baseline ankle torques. The peak resistance torque applied by the exosuit was 0.13 ± 0.01, 0.19 ± 0.01, 0.26 ± 0.02, and 0.32 ± 0.02 N m kg−1, in the LOW, MED, HIGH, and MAX bouts, respectively.


    (1) Across all bouts, participants increased peak ipsilateral biological ankle torque by 0.13–0.25 N m kg−1(p < 0.001) during exosuit-applied resistance compared to corresponding baselines. Additionally, ipsilateral soleus activity during stance increased by 5.4–11.3% (p < 0.05) in all but the LOW bout. (2) In the HIGH and MAX bouts, vertical ground reaction force decreased on the ipsilateral limb while increasing on the contralateral limb (p < 0.01). Secondary analysis found that the force magnitude that maximized increases in biological ankle torque without significant changes in limb loading varied by subject. (3) Finally, peak ipsilateral plantarflexion angle increased significantly during post-exposure in the intermediate HIGH resistance bout (p < 0.05), which corresponded to the greatest average increase in soleus activity (p > 0.10).


    Targeted resistance of ankle plantarflexion during stance by an exosuit consistently increased local ipsilateral plantarflexor effort during active resistance, but force magnitude will be an important parameter to tune for minimizing the involvement of the unresisted joints and limb during training.

    more » « less