Planned intervention: On Wednesday April 3rd 05:30 UTC Zenodo will be unavailable for up to 2-10 minutes to perform a storage cluster upgrade.
Published March 21, 2019 | Version v1
Presentation Open

Comparison of EMG pattern separability in the affected and non-affected arm in individuals with amputation

Description

Comparison of EMG pattern distinctness in the affected and non-affected arm in amputees

Authors:           A.W. Franzke, M.B. Kristoffersen, A. Murgia, R.M. Bongers, C.K. van der Sluis

Presenter:       Andreas W. Franzke

Affiliation:       Univ. of Groningen, Univ. Medical Center Groningen, The Netherlands

E-mail:             a.w.franzke@umcg.nl

 

Abstract

In myocontrol of upper limb prostheses using machine learning techniques, a basic requirement is the user’s ability to generate sufficiently distinct surface electromyography (sEMG) signals for different movement intents, over a wide range of arm orientations.1 Experiments with regard to training this ability are often conducted on able bodied, but it is unclear to what extent findings in non-affected limbs are representative for the affected limb. In this study we investigated whether sEMG patterns are more distinct in the unaffected compared to the affected side, and whether distinctness is affected by arm posture. 11 individuals with transradial amputation performed seven bimanual movements in three different arm orientations (hanging down; on arm rest; reaching out in front). sEMG patterns were recorded simultaneously on both arms with 8 electrodes. Distinctness was assessed by estimating the distance between sEMG patterns in the sEMG feature space using a modified mahalanobis distance measure. The data analysis showed that distinctness was significantly higher in the unaffected arm (p = 0.027) and no effect for orientation of arm and no interaction effect between arm and orientation of arm (p = 0.21) was found. These findings suggest that sEMG pattern distinctness is negatively affected by the absence of a limb. Generating sEMG patterns of sufficient distinctness for proper myoelectric control might therefore be more challenging for an individual with amputation compared to an able-bodied individual, which could have implications for machine learning myocontrol studies conducted on able bodied populations.

 

Acknowledgement

This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 687795, project Acronym INPUT. The content of 
this presentation does not reflect the official opinion of the European Union. Responsibility for the information and views expressed in therein lies entirely with the authors.

 

References

1. Resnik L, Huang H, Winslow A, et al. Evaluation of EMG pattern recognition for upper limb prosthesis control: a case study in comparison with direct myoelectric control. J Neuroeng Rehabil 2018; 15: 23.

 

Presented during TIPS/ISPO/BACPAR "Moving Beyond The Lab" on 20-23 March 2019 in Salford, Manchester.

Files

TIPS2019_Franzke.pdf

Files (1.2 MB)

Name Size Download all
md5:7de47bcf19e9984b72aa7ae77f350422
1.2 MB Preview Download

Additional details

Funding

INPUT – Intuitive Natural Prosthesis UTilization 687795
European Commission