Limits...
Control of Leg Movements Driven by EMG Activity of Shoulder Muscles.

La Scaleia V, Sylos-Labini F, Hoellinger T, Wang L, Cheron G, Lacquaniti F, Ivanenko YP - Front Hum Neurosci (2014)

Bottom Line: The temporal structure of the burst-like EMG activity was used to predict the spatiotemporal kinematic pattern of the forthcoming step.A comparison of actual and predicted stride leg kinematics showed a high degree of correspondence (r > 0.9).The proposed approach may have important implications for the design of human-machine interfaces and neuroprosthetic technologies such as those of assistive lower limb exoskeletons.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Neuromotor Physiology, Santa Lucia Foundation , Rome , Italy ; Centre of Space Bio-Medicine, University of Rome Tor Vergata , Rome , Italy.

ABSTRACT
During human walking, there exists a functional neural coupling between arms and legs, and between cervical and lumbosacral pattern generators. Here, we present a novel approach for associating the electromyographic (EMG) activity from upper limb muscles with leg kinematics. Our methodology takes advantage of the high involvement of shoulder muscles in most locomotor-related movements and of the natural co-ordination between arms and legs. Nine healthy subjects were asked to walk at different constant and variable speeds (3-5 km/h), while EMG activity of shoulder (deltoid) muscles and the kinematics of walking were recorded. To ensure a high level of EMG activity in deltoid, the subjects performed slightly larger arm swinging than they usually do. The temporal structure of the burst-like EMG activity was used to predict the spatiotemporal kinematic pattern of the forthcoming step. A comparison of actual and predicted stride leg kinematics showed a high degree of correspondence (r > 0.9). This algorithm has been also implemented in pilot experiments for controlling avatar walking in a virtual reality setup and an exoskeleton during over-ground stepping. The proposed approach may have important implications for the design of human-machine interfaces and neuroprosthetic technologies such as those of assistive lower limb exoskeletons.

No MeSH data available.


Shoulder muscle EMG-based prediction of stepping kinematics. (A) Schematic algorithm. Upper traces – actual left and right shank elevation angles during three consecutive strides. Lower traces – rectified (gray) and low-pass filtered (2 Hz, black) EMGs of shoulder muscles. The algorithm consisted in searching the EMG peak (τi) at the end of each step (ti) during the appropriate time window [(τi−1 + Δ, ti), see insert] that exceeded the pre-defined individually adjusted threshold (green line). Each step duration (Ti, Ti+1, etc.,) was predicted from the timing of the shoulder EMG peaks (Ti = τi − τi−1). (B) Predicted kinematic patterns.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4202724&req=5

Figure 1: Shoulder muscle EMG-based prediction of stepping kinematics. (A) Schematic algorithm. Upper traces – actual left and right shank elevation angles during three consecutive strides. Lower traces – rectified (gray) and low-pass filtered (2 Hz, black) EMGs of shoulder muscles. The algorithm consisted in searching the EMG peak (τi) at the end of each step (ti) during the appropriate time window [(τi−1 + Δ, ti), see insert] that exceeded the pre-defined individually adjusted threshold (green line). Each step duration (Ti, Ti+1, etc.,) was predicted from the timing of the shoulder EMG peaks (Ti = τi − τi−1). (B) Predicted kinematic patterns.

Mentions: Our approach uses the timing of the burst-like EMG activity of shoulder muscles (by applying the peak detection algorithm) to predict the spatiotemporal kinematic pattern of the forthcoming step. Prior to application of the peak detection algorithm, the EMG data were pre-processed: high-pass filtered at 30 Hz, rectified, and finally low-pass filtered (all filters, zero-lag fourth order Butterworth). Low-pass filtering was performed at different frequencies (1 ÷ 5 Hz) to achieve the best correlation between actual and predicted leg kinematics. Despite some inter-individual variability, periods of EMG activity of DELTa and DELTp during normal walking tend to be alternating and correspond to those of the contralateral upper limb (Ivanenko et al., 2006; Kuhtz-Buschbeck and Jing, 2012), as well as multi-muscle synergy-based control interface may be more efficient than a single-muscle control (Lunardini et al., 2014). Therefore, bilateral EMGs of synergistic muscles (Ivanenko et al., 2006; Kuhtz-Buschbeck and Jing, 2012) were summed (Figure 1A):(1)EMG1=DELTaright+DELTpleft(2)EMG2=DELTaleft+DELTprightthe peaks of EMG1 and EMG2 occur around the beginning of the swing phase of the right and left legs, respectively (Figure 1A). Nevertheless, we also compared the performance of our algorithm using all four muscles (Eqs 1 and 2) and only pairs of contralateral muscles (DELTaleft and DELTaright; DELTpleft and DELTpright).


Control of Leg Movements Driven by EMG Activity of Shoulder Muscles.

La Scaleia V, Sylos-Labini F, Hoellinger T, Wang L, Cheron G, Lacquaniti F, Ivanenko YP - Front Hum Neurosci (2014)

Shoulder muscle EMG-based prediction of stepping kinematics. (A) Schematic algorithm. Upper traces – actual left and right shank elevation angles during three consecutive strides. Lower traces – rectified (gray) and low-pass filtered (2 Hz, black) EMGs of shoulder muscles. The algorithm consisted in searching the EMG peak (τi) at the end of each step (ti) during the appropriate time window [(τi−1 + Δ, ti), see insert] that exceeded the pre-defined individually adjusted threshold (green line). Each step duration (Ti, Ti+1, etc.,) was predicted from the timing of the shoulder EMG peaks (Ti = τi − τi−1). (B) Predicted kinematic patterns.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4202724&req=5

Figure 1: Shoulder muscle EMG-based prediction of stepping kinematics. (A) Schematic algorithm. Upper traces – actual left and right shank elevation angles during three consecutive strides. Lower traces – rectified (gray) and low-pass filtered (2 Hz, black) EMGs of shoulder muscles. The algorithm consisted in searching the EMG peak (τi) at the end of each step (ti) during the appropriate time window [(τi−1 + Δ, ti), see insert] that exceeded the pre-defined individually adjusted threshold (green line). Each step duration (Ti, Ti+1, etc.,) was predicted from the timing of the shoulder EMG peaks (Ti = τi − τi−1). (B) Predicted kinematic patterns.
Mentions: Our approach uses the timing of the burst-like EMG activity of shoulder muscles (by applying the peak detection algorithm) to predict the spatiotemporal kinematic pattern of the forthcoming step. Prior to application of the peak detection algorithm, the EMG data were pre-processed: high-pass filtered at 30 Hz, rectified, and finally low-pass filtered (all filters, zero-lag fourth order Butterworth). Low-pass filtering was performed at different frequencies (1 ÷ 5 Hz) to achieve the best correlation between actual and predicted leg kinematics. Despite some inter-individual variability, periods of EMG activity of DELTa and DELTp during normal walking tend to be alternating and correspond to those of the contralateral upper limb (Ivanenko et al., 2006; Kuhtz-Buschbeck and Jing, 2012), as well as multi-muscle synergy-based control interface may be more efficient than a single-muscle control (Lunardini et al., 2014). Therefore, bilateral EMGs of synergistic muscles (Ivanenko et al., 2006; Kuhtz-Buschbeck and Jing, 2012) were summed (Figure 1A):(1)EMG1=DELTaright+DELTpleft(2)EMG2=DELTaleft+DELTprightthe peaks of EMG1 and EMG2 occur around the beginning of the swing phase of the right and left legs, respectively (Figure 1A). Nevertheless, we also compared the performance of our algorithm using all four muscles (Eqs 1 and 2) and only pairs of contralateral muscles (DELTaleft and DELTaright; DELTpleft and DELTpright).

Bottom Line: The temporal structure of the burst-like EMG activity was used to predict the spatiotemporal kinematic pattern of the forthcoming step.A comparison of actual and predicted stride leg kinematics showed a high degree of correspondence (r > 0.9).The proposed approach may have important implications for the design of human-machine interfaces and neuroprosthetic technologies such as those of assistive lower limb exoskeletons.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Neuromotor Physiology, Santa Lucia Foundation , Rome , Italy ; Centre of Space Bio-Medicine, University of Rome Tor Vergata , Rome , Italy.

ABSTRACT
During human walking, there exists a functional neural coupling between arms and legs, and between cervical and lumbosacral pattern generators. Here, we present a novel approach for associating the electromyographic (EMG) activity from upper limb muscles with leg kinematics. Our methodology takes advantage of the high involvement of shoulder muscles in most locomotor-related movements and of the natural co-ordination between arms and legs. Nine healthy subjects were asked to walk at different constant and variable speeds (3-5 km/h), while EMG activity of shoulder (deltoid) muscles and the kinematics of walking were recorded. To ensure a high level of EMG activity in deltoid, the subjects performed slightly larger arm swinging than they usually do. The temporal structure of the burst-like EMG activity was used to predict the spatiotemporal kinematic pattern of the forthcoming step. A comparison of actual and predicted stride leg kinematics showed a high degree of correspondence (r > 0.9). This algorithm has been also implemented in pilot experiments for controlling avatar walking in a virtual reality setup and an exoskeleton during over-ground stepping. The proposed approach may have important implications for the design of human-machine interfaces and neuroprosthetic technologies such as those of assistive lower limb exoskeletons.

No MeSH data available.