Decoding motor neuron behavior for advanced control of upper limb prostheses

by Tamás Kapelner
Doctoral thesis
Date of Examination:2016-12-01
Date of issue:2016-12-16
Advisor:Prof. Dr. Dario Farina
Referee:Prof. Dr. Dario Farina
Referee:PD Dr. David Liebetanz
Persistent Address: http://dx.doi.org/10.53846/goediss-6007

 

 

Files in this item

Name:Dissertation_TK_fastwebview.pdf
Size:11.7Mb
Format:PDF
Description:Dissertation

The following license files are associated with this item:

Abstract

English

One of the main challenges in upper limb prosthesis control to date is to provide devices intuitive to use and capable to reproduce the natural movements of the arm and hand. One approach to solve this challenge is to use the same control signals for prosthesis control that our nervous system uses to control its muscles. This thesis aims to investigate the possibility of natural, intuitive prosthesis control using neural information obtained with available surface EMG decomposition methods. In order to explore all aspects of such a novel approach, a series of five studies were performed with the final goal of implementing a proof of concept and comparing its performance with state of the art myoelectric control. The performed investigations revealed important insights in motor unit physiology after targeted muscle reinnervation, EMG decomposition in dynamic voluntary contractions of the forearm, and the properties and challenges of neural information based prosthesis control. The main outcome of the thesis is that neural information based prosthesis control is capable to outperform myoelectric approaches in pattern recognition, linear regression and nonlinear regression, as determined by offline performance comparisons. The final proof of concept for this novel approach was a robust regression method based on neuromusculoskeletal modeling. The kinematics estimation of the proposed approach outperformed EMG-based nonlinear regression in both able-bodied subjects and patients with limb deficiency, indicating that using neural information is a promising avenue for advanced myoelectric control.
Keywords: EMG; Prosthesis control; Neural information; EMG decomposition; TMR
 
Statistik