Electrode–skin impedance compensation for improved bioelectrical signal acquisition (2020)
Type of ContentTheses / Dissertations
Thesis DisciplineMechanical Engineering
Degree NameDoctor of Philosophy
PublisherUniversity of Canterbury
Stroke is the third leading cause of disability worldwide, commonly removing a subject’s independence. However, physical therapy can assist a subject in regaining their lost functionality. Modern physical therapy is incorporating assistive robotic devices, allowing more intensive and repetitive training while reducing therapy cost.
The incorporation of surface electromyography (sEMG) into assistive robotics can enable patient-driven intention-based control, leading to increased patient interaction and a more natural, unconscious interface. sEMG is the non-invasive technique of measuring the bio- electrical activity of the skeletal muscle at the skin surface. The bioelectrical signal is bipolar with an amplitude of ≤ 10 mVpk–pk, a frequency spectrum of 0–500 Hz and is typically measured using two recording electrodes and a single reference electrode. However, electrical interference and bioelectrical crosstalk limit the efficacy of bioelectrical feedback for assistive robotic control.
In built-up environments, the human body is capacitively coupled to the mains power sup- ply and ground, leading to interference potentials which are a function of the impedance imbalance between recording electrodes. The current methods of reducing electrical interference either removes a portion of the signal of interest; can have limited affect, resulting in large interference potentials; or can be time consuming with the potential to lead to skin irritation.
Bioelectrical crosstalk, detected with sEMG, is the phenomenon of one muscle’s signal influencing the recording of another. Crosstalk can lead to misrepresentation of the target signal, increasing the difficulty to provide accurate biofeedback. The tripolar electrode configuration is commonly used to reduce crosstalk. However, using three electrodes increases the possibility of impedance imbalances between recording electrodes.
Previous research has focused on balancing the common-mode input impedance of the bioelectrical instrumentation device with the impedance of the electrode-skin interface. The common-mode interference potential was used as an indicator to control the required common-mode input impedance. However, without measuring the electrode-skin impedance, the unique transfer function of each electrode-skin interface will be unknown, reducing the ability to use the frequency spectrum of the bioelectrical signal for biofeed-back. Therefore, there is a need to balance the impedance between electrode-skin interfaces and determine the resulting transfer function between the electrode-skin interface and the bioelectrical instrumentation device.
Commercial, research-level sEMG devices do not have an open source signal processing architecture. Therefore, to quantify the impact of balancing the impedance of multiple electrode-skin interfaces, a high quality near-raw signal output sEMG device was developed. The device has a resolution of 298 nV, a baseline noise of less than 5.2 µVrms (when abrasive skin preparation is applied), a signal bandwidth of 21.2–433 Hz, a sampling rate of 1 kHz and built in 1 x 10 mm Ag bar electrodes.
Characterising the behaviour of the electrode-skin interface provides quantitative insight into optimising a compensatory system to balance the impedance between multiple electrode- skin interfaces. A method to model the step response of the electrode-skin interface was developed using simulation and physical, passive circuitry, resulting in a mean error per modelled component of 0.076% and 3.49% for simulation and passive circuitry, respectively.