Graduation Semester and Year
2019
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Biomedical Engineering
Department
Bioengineering
First Advisor
Khosrow Behbehani
Abstract
Continuous and non-invasive monitoring of blood pressure (BP) and cerebral blood flow (CBF) has numerous clinical and health maintenance applications. Of particular interest is monitoring of BP and CBF during events that stress the cardiovascular system such as cessation of breathing during sleep apnea. Current means and methods of continuous measurement of these signals are either invasive or require expensive equipment. Therefore, investigation of new methods of continuous estimation of the BP and CBF are of high interest. One attractive alternative is to use photoplethysmography (PPG) technique to estimate BP and CBF, as PPG is shown to be able to detect blood volume changes in peripheral vasculature. This study aims to investigate the feasibility of estimating key parameters of BP and CBF using PPG. Specifically, it is hypothesized that apnea-induced fluctuations in blood pressure and cerebral blood flow can be estimated using photoplethysmography. We propose a new method for the continuous estimation of systolic and diastolic values as well as peaks and troughs of cerebral blood flow. The estimates are obtained from a photoplethysmography signal as input to a fifth order autoregressive moving average (ARMA) models. The ARMA models are suited for physiological modeling because they are dynamic. They integrate past and present samples of input as well as past samples of the estimated output to estimate the present value of the output. Further, ARMA models accommodate pure time delays that are often present in physiological system response. To test the said hypothesis using the proposed ARMA modeling approach, we performed two studies: 1) breath-hold (BH) maneuver was used as a means of simulating apnea to elicit changes in both BP and CBF, and 2) polysomnography sleep (PSG) study of patients suspected of having obstructive sleep apnea (OSA) was conducted. During both studies, concurrent measurement of BP, CBF, and PPG was made. BH protocol consisted of 5 consecutive BH with 90 seconds normal breathing (NB) interval in between and 60 seconds of NB at the beginning and at the end. For BH study, beat-to-beat full-wave BP and CBF of 15 subjects (8 males, 7 females, aged 28.9 ± 5.0 years, BMI 24.1 ± 4.8 kg/m2) with no known cardiovascular disorder were recorded. For sleep study, same measurements of 15 OSA patients (10 males, 5 females, aged 53.8 ± 7.4 years, BMI 34.2 ± 7.2 kg/m2) were recorded during PSG study. From each subject’s dataset 6 longest apneas were selected to keep the statistical analysis as consistent as possible to the BH study. Each BH/NB interval, or apnea episode in case of sleep study, was evaluated with all other congruent intervals to ascertain the accuracy of the models and resulted in both modeling and validation errors which were further used to calculate means, standard deviations and rMSE. For BH study, the mean of model residuals for was less than 3 mmHg for BP modeling and less than 4 cm/s for CBF modeling and the root mean squared (rMSE) of the ARMA model residuals, was less than 8 mmHg during the study for modeling BP and less than 11 cm/s for CBF. Also, maximum standard deviation was calculated to be less than 9 mmHG for the BP modeling and less than 13 cm/s for CBF modeling. For the sleep apnea study, mean of residuals for BP was less than 8 mmHG and the rMSE of the model residuals was less than 22 mmHG. Further, for this study, the mean of residuals for CBF estimates was less than 13 cm/s and their rMSE values was less than 15 cm/s. Standard deviation of the errors were all less than 34 mmHG with average of 15.83 mmHg for BP modeling and 20 cm/s with average of 7.28 cm/s for CBF modeling. BH modeling results indicate that ARMA model can provide estimates with adequate accuracy to be of practical use in estimating BP and CBF values. In contrast, for PSG study, further development, including better feature selection and possibly non-linear modeling may be needed to increase the accuracy of the estimates. Results of both studies suggest that with further development, it may be possible to get reasonable estimates of key features of blood pressure or cerebral blood flow from PPG, creating opportunities for improvement in monitoring cerebrovascular and cardiovascular system health. One such opportunity is the ability to nocturnally measure blood pressure and cerebral blood flow using now widely available and reasonably-priced pulse oximeter systems, as they are capable of delivering PPG signal.
Keywords
Blood pressure, Cerebral blood flow, Modeling, Autoregressive, Photoplethysmography, Apnea, Sleep
Disciplines
Biomedical Engineering and Bioengineering | Engineering
License
This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.
Recommended Citation
Soltan Zadi, Armin, "MATHEMATICAL MODELING OF BLOOD PRESSURE AND CEREBRAL BLOOD FLOW USING PHOTOPLETHYSMOGRAPHY SIGNAL" (2019). Bioengineering Dissertations. 135.
https://mavmatrix.uta.edu/bioengineering_dissertations/135
Comments
Degree granted by The University of Texas at Arlington