A comparative study of myosignals taken from constrained and non-constrained hand and wrist movements

Abstract

Introduction: despite the perceived benefits of myoelectric prostheses, rejection rates remain high. Key issues include the artifacts and delay of the myosignals used for myoelectric control [1]. When researching myosignals it is common to substitute amputee data with that of an intact limb. To facilitate this, it is important to study the effects on myoelectric signals during the constraint of the wrist, hand, and forearm with a view to establishing reliable myosignal features that the control techniques rely upon. Although prosthetic hand simulators have been reported [2], there lacks a systematic study to provide better insight underpinning intelligent control development.
This study compares constrained and non-constrained myosignals of a volunteer’s intact limb using a purposely designed hand-wrist cuff.

Methods: an Ottobock MyoBoy system, a clinically applicable amputee training tool, was connected to an Ottobock MyoHand VariPlus Speed prosthetic hand. Two MyoBoy electrodes were applied at the flexor carpi ulnaris and extensor carpi ulnaris of the volunteer’s intact forearm. The volunteer was asked performed eight consecutive muscle activations, alternating between the flexor carpi ulnaris and extensor carpi ulnaris, aiming to open and close the prosthetic hand. This was achieved through wrist flexions and extensions by the volunteer. Two distinct activity sets, i.e., fast prosthetic hand activations and slow prosthetic hand activations, were performed to mimic clinical training protocol. These activities were conducted in both constrained and non-constrained conditions, where constraint was achieved through a purposely designed and 3D printed hand and wrist cuff with bespoke fit for the volunteer. This study was then repeated at different electrode gain settings.

Results & Discussion: significant differences in myosignal quality were observed when removing limb articulation. During non-constrained fast activation, an average peak myosignal of 145 was observed during flexion, over 70% greater than its constrained counterpart of 39, and an average peak myosignal of 167 was observed during extension, over 60% greater than its constrained counterpart of 54. Similar trend was also observed for the average peak myosignals during slow activation. As a result, non-constrained tests resulted in full prosthetic articulation success rates of over 95%. In contrast, these values are significantly lower for the constrained counterparts, i.e., gave full prosthetic articulation success rates under 5%. Further details, including the effect on the myosignal gain settings, will also be presented and its potential influence in advanced feature detection and control development will be discussed.

Conclusion: this study provides quantitative analysis on myosignal variation under hand constraint and non-constraint conditions, during simulated tests in a lab setting. The results suggest the importance of restricting hand and wrist articulation to allow validated myosignals for research e.g. development of advanced control strategies for intelligent prosthesis.

References
[1] Chadwell, A., Kenney, L., Thies, S., Galpin, A. and Head, J., 2016. The reality of myoelectric prostheses: understanding what makes these devices difficult for some users to control. Frontiers in neurorobotics, p.7.

[2] Sinke, M., Chadwell, A. and Smit, G., 2022. State of the art of prosthesis simulators for the upper limb: A narrative review. Annals of Physical and Rehabilitation Medicine, 65(6), p.101635.

More information

Identifiers

ORCID for K.M. Devin: orcid.org/0000-0001-6794-2375

ORCID for J. Tang: orcid.org/0000-0003-3359-5891

ORCID for L. Jiang: orcid.org/0000-0002-3400-825X

Catalogue record

Export record