Signal Quality Analysis in Pulse Oximetry:
Modelling and Detection of Motion Artifact
by
Geoffrey Clarke, B.Eng.
A thesis submitted to the
Faculty of Graduate and Postdoctoral Affairs
in partial fulfillment of the requirements for the degree of
Master of Applied Science in Biomedical Engineering
Ottawa-Carleton Institute for Biomedical Engineering
Department of Systems and Computer Engineering
Carleton University
Ottawa, Ontario
May 2015
c
Copyright
Geoffrey Clarke, 2015
The undersigned hereby recommends to the
Faculty of Graduate and Postdoctoral Affairs
acceptance of the thesis
Signal Quality Analysis in Pulse Oximetry: Modelling and
Detection of Motion Artifact
submitted by Geoffrey Clarke, B.Eng.
in partial fulfillment of the requirements for the degree of
Master of Applied Science in Biomedical Engineering
Professor Adrian D.C. Chan, Thesis Co-Supervisor
Professor Andy Adler, Thesis Co-supervisor
Professor Roshdy Hafez, Chair,
Department of Systems and Computer Engineering
Ottawa-Carleton Institute for Biomedical Engineering
Department of Systems and Computer Engineering
Carleton University
May, 2015
ii
Abstract
Pulse oximetry is a non-invasive technique for measuring the amount of oxygen in
a patient’s arterial blood, as a percentage of the blood’s oxygen carrying capacity
(SpO
2
). This measurement is considered standard of care in the hospital for mon-
itoring the cardio-respiratory function of a patient. While it has potential uses in
ambulatory or wearable monitoring applications, pulse oximetry is particularly sus-
ceptible to motion artifact contamination. This thesis presents efforts to quantify and
model the effects of motion artifact, and automatically detect periods of poor signal
quality.
First, the effects of motion artifact on SpO
2
are analyzed using motion contam-
inated data. Second, two models are identified from previous literature that may
explain the effects of motion artifact on pulse oximetry. These models are developed
analytically and evaluated using isolated motion artifact signals. Finally, three auto-
matic signal quality assessment algorithms are proposed. These algorithms are shown
to discriminate between clean and motion contaminated signals.
Overall, this thesis attempts to inform the development of software and hardware
based techniques to mitigate the effects of poor signal quality on pulse oximetry.
iii
Acknowledgments
I would like to express my gratitude to my supervisors, Dr. Adrian Chan and Dr.
Andy Adler for their continuous support of my studies. Their expertise, mentorship
and patience provided for a fantastic graduate experience.
To my colleagues Patrick Quesnel and Chris Andison, thank you for your brain-
storming, debates, and editorial assistance. I look forward to working with you in
the future. Thanks to Kait Duncan for your support and editorial assistance, and
for making me dinner sometimes, even though you’re busier than me. Despite your
patience, I’m sure you’re as happy as I am that I’m done!
Finally, I am grateful to my family, who have always taken interest in my academic
progress. Thank you for your unwavering support, patience, and encouragement to
take on this endeavour in the first place.
iv
Statement of Originality
This thesis presents the work of the author, under the supervision of Dr. Adrian
D. C. Chan and Dr. Andy Adler. This work was completed at Carleton University
for the degree Master of Applied Science in Biomedical Engineering. Some of these
results have been or will be presented in conference publications:
1. G. W. J. Clarke, A. D. C. Chan, and A. Adler, “Effects of motion artifact on the
blood oxygen saturation estimate in pulse oximetry,” 2014 IEEE International
Symposium on Medical Measurements and Applications (MeMeA), pp. 14, 2014.
• This conference paper describes the initial work in quantifying the effect of
motion artifact in pulse oximetry. This work was expanded upon in Chap-
ter 4, using an improved blood-oxygen saturation calculation algorithm,
and comparing to a contaminant free reference signal recorded in parallel.
This paper was presented by the author at the 2014 IEEE International
Symposium Medical Measurements and Applications (MeMeA).
2. G. W. J. Clarke, A. D. C. Chan, and A. Adler, “Quantifying Blood-Oxygen Sat-
uration Measurement Error in Motion Contaminated Pulse Oximetry Signals,”
World Congress on Medical Physics and Biomedical Engineering, 2015.
• This conference paper describes the collection of isolated motion artifact
signals, allowing for artificial signal contamination with precise control over
v