10.5281/zenodo.3614902
https://zenodo.org/records/3614902
oai:zenodo.org:3614902
Franzke, AW
AW
Franzke
Kristoffersen, MB
MB
Kristoffersen
van der Sluis, CK
CK
van der Sluis
Bongers, RM
RM
Bongers
Murgia, A
A
Murgia
Myoelectric Assitive Devices: Does EMG distinctness reflect control ability
Zenodo
2018
Myoelectric Control
Prosthetics
Pattern Recognition
2018-12-12
eng
Presentation
10.5281/zenodo.3614901
https://zenodo.org/communities/eu
Creative Commons Attribution 4.0 International
MYOELECTRIC ASSISTIVE DEVICES: DOES EMG PATTERN DISTINCTNESS REFLECT CONTROL ABILITY?
Authors: A.W. Franzke1, M. B. Kristoffersen1, C. K. van der Sluis1, R.M. Bongers2, A. Murgia2,
1 University of Groningen, University Medical Center Groningen, Department of Rehabilitation Medicine, Groningen, the Netherlands
2 University of Groningen, University Medical Center Groningen, Center for Human Movement Sciences, Groningen, the Netherlands
Email corresponding author: a.w.franzke@umcg.nl
Purpose:The literature suggests that improvements in control of a pattern-recognition based myoelectric device are governed by increased distinctness of surface EMG patterns (dEMG). We investigated the relation between control ability and dEMG.
Methods: Able-bodied participants learned to control a pattern-recognition based myoelectric device over 5 days. Each day, they were fitted with 8 surface EMG electrodes around their forearm and performed 3 training sessions. The procedure in each session was as follows: First, participants performed muscle contractions following movement prompts (e.g. “wrist flexion”). The EMG signals were used to train a software algorithm to recognize the different EMG patterns associated to each movement. Second, a test was performed where the same movement prompts were presented to the participants again, who performed muscle contractions accordingly. The algorithm then estimated the movement performed by the participant.
Performance was assessed by the percentage of correctly estimated movements (pCEM) per session. dEMG was calculated as a modified Mahalanobis distance in the EMG feature space per session. The strength of the relationship between pCEM and dEMG was assessed by fitting a function to the data (across all participants and sessions) and calculating the goodness of fit.
Results: 33 participants (mean age 21.6ys, 17 females) completed the study. From first to last session mean pCEM increased from 64% to 85% (p <.0001, η = 0.43). An exponential fit of dEMG to pCEM ( ) resulted in R2 = 0.33 (with a standard error of estimate of 0.25).
Conclusions: Performance significantly increased from first to last session, but the correlation between dEMG and pCEM was poor. This suggests that participants could improve their performance without making EMG patterns more distinct, indicating that the ability to control a pattern-recognition based myoelectric device is not strongly reflected in EMG pattern distinctness.
Presented during the 6th RehabMove Congress on December 12-14, 2018 in Groningen, The Netherlands.
European Commission
10.13039/501100000780
687795
Intuitive Natural Prosthesis UTilization