Date of Award

January 2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical Engineering

First Advisor

Kouhyar Tavakolian

Abstract

Respiration affects the cardiovascular system significantly and the morphology of signals relevant to the heart changes with respiration. Such changes have been used to extract respiration signal from electrocardiogram (ECG). It is also shown that accelerometers placed on the body can be used to extract respiration signals. It has been demonstrated that the signal morphology for seismocardiogram, the lower frequency band of chest accelerations, is different between inhale and exhale. For instance, systolic time intervals (STI), which provide a quantitative estimation of left ventricular performance, vary between inhale and exhale phases. In other words, heart beats happening in exhale phase are different compared to those in inhale phase. Thus, our main goal in this thesis is investigating feasibility of finding an automatic morphological based method to identify respiratory phases of heart cycles.

In this thesis, forty signal recordings from twenty subjects were used. In each recording, the reference respiratory belt signal, three dimensional (3D) chest acceleration signals, and electrocardiogram signals were recorded. The first stage was is choosing a proper estimated respiratory signal. The second stage, was the automatic respiratory phase detection of heart cycles using the selected estimated respiratory signal. The result shows that among estimated respiratory signals, accelerometer-derived respiration (ADR), in z-direction, has a potential m to identify respiratory phase of heart cycles with total accuracy of about 77%.

Share

COinS