Insulated Electromyography for Myoelectric Prosthesis Control

Myoelectric prostheses help amputees to regain independence and a higher quality of life. Such prostheses are controlled by the remaining muscle tissue. An electrical signal can be measured at the surface of the skin, the so-called electromyography signal. State-of-the-art sensors use a conductive connection to the skin, which is sensitive to sweat. These sensors cause pressure marks, which are detrimental for patients with circulatory disorders, however, these constitute a major group of the amputees. In this thesis, insulated electromyography sensors, with an insulating layer between skin and sensor electrode, are suggested instead. These sensors are insensitive to sweat and due to their flexible design, they do not cause pressure marks. Especially textile sensors are comfortable and they can be sewed into the prosthetic shaft or cloth. The optimal sensor geometry was determined based on measurements and the simulated electrical potential at the skin surface. Different sensor assemblies were investigated to achieve high signal coupling. A special focus was given to the robustness against motion artifacts, which constitute a major obstacle in biosignal measurement. These are caused by external mechanical strikes, lifting or shifting of the sensor or the relative movement of the muscle to the skin. The prosthesis drive should not be activated at motion artifacts or due to other noise, such as the power-line interference. The insulated sensor assembly, the analog circuit and the digital signal processing were designed to be robust in real-world environment, low-power and compact. Insulated electromyography requires precision electronics and optimal filter parameters for high signal quality. Different shielding techniques were evaluated to protect the signal from external noise and parasitic capacities, which would attenuate the signal amplitude. To achieve long battery lifetime in the mobile system, low-power components and algorithms were implemented. Algorithms, which distinguish between artifacts and contraction electromyography signal were implemented. The remaining capacity due to the short runtime of the digital signal processing allows for the implementation of additional algorithms to further increase the robustness of the system.submitted by DI Theresa RolandUniversität Linz, Dissertation, 2019OeBB(VLID)366368