The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	Motion-resistant pulse oximetry
Author:	Alkhoury, Ludvik
View Online:	njit-etd2023-017 (xxvii, 114 pages ~ 25.5 MB pdf)
Department:	Department of Electrical and Computer Engineering
Degree:	Doctor of Philosophy
Program:	Electrical Engineering
Document Type:	Dissertation
Advisory Committee:	Kam, Moshe (Committee chair) Hou, Edwin (Committee member) Ansari, Nirwan (Committee member) Abdi, Ali (Committee member) Liu, Xuan (Committee member) Pal, Saikat (Committee member)
Date:	2023-05
Keywords:	Digital signal processing Health monitoring systems Motion-resistant algorithms Multi-wavelength oximetry Vital signs calculation Wearable devices
Availability:	Unrestricted
Abstract:	The measurement of vital signs — such as peripheral capillary oxygen saturation (SpO2) and heart rate (HR) levels — by a pulse oximeter is studied. The pulse oximeter is a non-invasive device that measures photoplethysmography (PPG) signals and extracts vital signs from them. However, the quality of the PPG signal measured by oximetry sensors is known to deteriorate in the presence of substantial human and sensor movements contributing to the measurement noise. Methods to suppress such noise from PPG signals measured by an oximeter and to calculate the associated vital signs with high accuracy even when the wearer is under substantial motion are presented in this study. The spectral components of the PPG waveform are known to appear at a fundamental frequency that corresponds to the participant's HR and at its harmonics. To match this signal, a time-varying comb filter tuned to the participant's HR is employed. The filter captures the HR components and eliminates most other artifacts. A significant improvement in the accuracy of SpO2 calculated from the comb-filtered PPG signals is observed, when tested on data collected from human participants while they are at rest and while they are exercising. In addition, an architecture that integrates SpO2 levels from multiple PPG channels mounted on different parts of the wearer's arm is presented. The SpO2 levels are integrated using a Kalman filter that uses past measurements and modeling of the SpO2 dynamics to attenuate the effect of the motion artifacts. Again, data collected from human participants while they are at rest and while they are exercising are used. The integrated SpO2 levels are shown to be more accurate and reliable than those calculated from individual channels. Motion-resistant algorithms typically require an additional noise reference signal to produce high quality vital signs such as HR. A framework that employs PPG sensors only — one in the green and one in the infrared spectrum — to compute high quality HR levels is developed. Our framework is tested on experimental data collected from human participants while at rest and while running at various speeds. Our "PPG-only" framework generates HR levels with high accuracy and low computational complexity as compared to leading HR calculation methods in the literature that require the availability of a noise reference signal. The methods for SpO2 and HR calculation presented in this study are desirable since (1) they yield high accuracy in estimating vital signs under substantial level of motion artifacts and (2) they are computationally efficient, (and therefore are capable to be implemented in wearable devices).