Optimizing an ICA-Wavelet Denoising Method for investigating EMGdi Reflexes

Author: Tianshu Chu

Chu, Tianshu, 2018 Optimizing an ICA-Wavelet Denoising Method for investigating EMGdi Reflexes, Flinders University, College of Science and Engineering

This electronic version is made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. This thesis may incorporate third party material which has been used by the author pursuant to Fair Dealing exceptions. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.

Abstract

Diaphragmatic electromyography (EMGdi) signals can be recorded from surface electrodes placed on the chest wall, intra-muscular electrodes placed directly into the muscle, or via multi-channel intra-oesophageal electrode recordings. EMGdi recordings contain detailed information regarding the central neural drive to breathe and mechanoreflex mediated changes in muscle electrical activity over the course of each breath, and well known to operate in other respiratory modulated pump muscles such as the scalene, and upper airway dilator muscles such as the genioglossus. Thus, assessment of EMGdi activity can help with the assessment of overall neural drive to breathe, and in exploring respiratory pathological mechanisms and respiratory reflex mechanisms. Although somewhat invasive, intra-oesphageal recordings provide high quality EMGdi recordings without contamination by intercostal muscle activity or the attendant risks of pneumothorax and infection associated with intramuscular recordings. However, raw EMGdi signals are heavily contaminated by ECG artefact, particularly when recorded via an intra-osephageal catheter. Thus, reliable assessment of respiratory related intra-esophageal EMGdi requires removal of ECG interference. Conventional methods for ECG denoising of EMGdi predominantly rely on simplistic ECG blanking methods that ignore EMGdi periods containing ECG artefact, or substitute artefact periods with delayed EMGdi recorded a few hundred milliseconds earlier within the respiratory cycle. Whilst these methods are adequate for assessing overall tonic and peak inspiratory levels of EMGdi activity (e.g. from rectified and moving averaged EMGdi after ECG blanking), they are not appropriate for examining within breath reflex changes in inspiratory activity in response to within-breath changes in inspiratory loads, such as mid-inspiratory occlusion. Examination of these reflexes requires signal averaging of raw rectified EMG over many replicated stimuli in order to sufficiently improve signal-to-noise to discern small EMG changes associated with these reflexes. Given that conventional ECG blanking methods destroy large segments of underlying EMGdi activity, signal averaging methods cannot reliably be applied. Averaging of raw unfiltered EMGdi inevitably remains

heavily contaminated with ECG. Consequently, meaningful examination of EMGdi reflex responses to mid-inspiratory occlusion has not previously been possible. Recently described ECG filtering methods offer significant promise, but have yet to be applied to examine EMGdi reflex responses. The purpose of the work described in this thesis was to apply newly described EMGdi filtering methods to test, for the first time, if this new approach can allow for assessment of EMGdi reflex responses to mid-inspiratory occlusion.

Keywords: Biomedical, ICA-Wavelet Denoising Method, EMGdi Reflexes

Subject: Engineering thesis

Thesis type: Masters
Completed: 2018
School: College of Science and Engineering
Supervisor: Karen Reynolds